US20080018532A1

US20080018532A1 - Monitoring sports and swimming

Info

Publication number: US20080018532A1
Application number: US11/843,204
Authority: US
Inventors: Colin Mackintosh; Daniel James; Neil Davey; Ronald Grenfell; Kefei Zhang
Original assignee: Sportzco Pty Ltd
Current assignee: Catapult Innovations Pty Ltd
Priority date: 2002-11-01
Filing date: 2007-08-22
Publication date: 2008-01-24
Also published as: US20060212271A1; ES2641232T3; EP1562680B1; WO2004039462A1; EP1562680A4; US7272499B2; EP1562680A1

Abstract

A data logger for a swimmer which includes an accelerometer, and a GPS unit to sense position and velocity, a heart rate monitor, a controller programmed to manipulate the data and provide a display of the heart rate, lap times, stroke rate etc. The data can be stored or transmitted to a remote computer for use by the coach. The device can also be adapted for other sports.

Description

This invention relates to a method and system for monitoring performance characteristics of swimmers and in particular the particular movements which contribute to enhanced performance.

BACKGROUND TO THE INVENTION

Monitoring of athletes performance both in training and in competition, is important in the development and implementation of new approaches aimed at improving sporting performance.
The ability to measure and record athlete physiological information and positional information associated with athlete movement in real-time is critical in the process of athlete training and coaching. Blood oxygen, respiration, heart rates, velocity, acceleration/force, changes in direction, and position and many other factors are required in elite athlete training and coaching. The position, movement and force information plays an important role in effective analysis of the athlete performance, especially for rowers. For example, the stroke frequency, force and synchronization of athletes are critical for the performance of the rowers in a competition. Currently the stroke information can only be measured in either dedicated sports laboratories or using simulated devices. Reliable analysis of the stroke rate and stroke distance in rowing has been a challenge for a long time due to the availability of the real scenario data, in particular a high precision of position, velocity and acceleration data. Existing technologies used for this purpose include theoretical studies, video-footage procedure, indoor tank procedure, computer modeling and ergometer studies. Much of the equipment is either too heavy, expensive, obtrusive or less reliable. Therefore, smart real-time monitoring during training and competition to help elite athletes to improve their performance and avoid injuries is critical for both athletes and coaches. Any methodology that would improve the situation would not only bring benefits to the rower practice, but also to many other sports related application including both team sports and individual athlete.
U.S. Pat. Nos. 4,984,986 and 5,099,689 disclose measuring systems for off water rowing apparatus which measure the number of strokes or the force applied to the machine.
U.S. Pat. No. 6,308,649 discloses a monitoring system for sail boat racing which provides feedback to the crew of such parameters as wind speed and direction boat speed, sail boat comfort parameters, sail shape, line tensions, rudder angle etc.
Some development of monitoring systems has occurred in non water sports. U.S. Pat. No. 6,148,262 discloses a bike mounted sports computer including a GPS receiver to provide a mapping facility.
U.S. Pat. No. 5,685,722 discloses swimmers goggles which incorporate a timer and a display. An accelerometer senses the tumble turns to count laps and the goggles include a strap to sense the pulse of the swimmer. Such head-mounted systems make it difficult to discriminate between the movements of the head and the rest of the body of the swimmer.
Wrist watch type sensors for swimmers have also been proposed as in U.S. Pat. No. 5,864,518 and application 2004/0020856.
Recently proposed swimming loggers prefer a display in the goggles.
U.S. Pat. No. 6,033,228 discloses a device attached to a swimmers waist. The device includes an impeller magnetic device which is able to signal changes in speed to a visual display worn by the swimmer.
U.S. Pat. No. 5,685,722 discloses a goggle mounted accelerometer and display.
WO/03061779 suggests displaying real time data visually in the goggles but does not suggest aural display. This disclosure favors separation of the display from the motion sensor which is located preferably on the back with RF transmission to the display. There is a suggestion of monitoring pulse rate using a temporal artery and to integrate the whole device into one unit on the goggles.
Accelerometers are able to detect changes in acceleration but do not provide a meaningful measure of velocity.
It is an object of this invention to provide a device for real time monitoring of swimmers that is useful during a training session and also for coaches to carry out detailed analysis after the training session.

BRIEF DESCRIPTION OF THE INVENTION

To this end the present invention provides a data acquisition system for use in swimming which incorporates

- a) a global position sensor to derive 3 dimensional positioning data relative to time elapse
- b) at least one accelerometer to derive acceleration and velocity data in 3 dimensions
- c) a microcontroller with a clock to interrogate the global position sensor preferably at a frequency of at least 1 Hz and to measure the accelerometer data
- d) a power supply
- e) communication means for transmission of global position and accelerometer data from the microcontroller to a remote computer device
- f) the remote computer device being programmed to use the global position and accelerometer data to provide accurate and continuous output of parameters such as velocity acceleration and distance traveled.

This device will provide positional data from the training and competition environment and provide both athlete physiological data and performance data related to the sport.
The movement sensor is an accelerometer combined with a GPS unit to sense instantaneous position and velocity. A GPS receiver transmitter is included in the device to derive location and speed parameters.
Preferably physiological sensors are also attached to the athlete and integrated with the sensor system. Heart rate is the prime parameter to be measured and this may be sensed using electrical sensors or microphones. Respiratory rate is also important and may be measured by sensing the stretching of a chest band or using a microphone and signal recognition software. Another parameter is arterial oxygen saturation which may be measured non invasively by a sensor, placed on an earlobe or finger tip, using pulse oximetry employing an infra red absorption technique. Infra-red spectroscopy may be used for non invasive measurement of blood lactate concentrations.
Preferably velocity is derived from the global position sensor and the accelerometer data is sampled to obtain movement characteristics of the sport being monitored. Preferably the accelerometer data is integrated to derive velocity related movement characteristics and drift is checked every second using the output from the global position sensor.
This system provides a platform device which can be used for a wide range of sports simply by providing appropriate software to derive from the accelerometer and GPS data, the desired sport parameters such as stride frequency velocity stride length, vertical acceleration, time off the ground for long jumping and events such as aerial skiing. The system of this invention can be used in swimming to identify stroke type, turns, and with turns the number of laps and the stroke rate per lap as well as lap times. Careful analysis of each stroke can show the efficiency and power by comparing the acceleration and deceleration cycles and the effect of breathing cycles. In open water swimming the GPS can also be used to provide an indication of location, direction, and speed relative to the course.
The device of this invention may also include an accelerometer so that tri-athletes who run and swim can obtain accelerometer (pedometer) based speed and distance data for the land portion of their activities. Similarly GPS devices may also be included to derive similar distance and speed data. By adding a magnetometer to the unit on the cyclist pedal cadence can be sensed.
Alternatively pedal cadence can be sensed by a unit on the bike and transferred to the unit on the cyclist using a radio frequency pickup unit.
The accelerometer information may also be used to determine stroke type, stroke count, turns, lap count, lap times and speed in swimming and stride count, stride length and speed in running and cadence and power in cycling.
In another embodiment the present invention provides a wrist mounted sensor with a large display screen able to communicate with a second unit mounted on the swimmers head. Because GPS may not receive signals when the device is submerged the GPS unit may be mounted in the unit on the head and communicates wirelessly with the device on the wrist when the wrist is out of the water. Alternatively it is within the scope of this invention to modify the GPS polling routine to ensure that basic location information can be received within the time interval that the wrist is clear of the water.
This means that the display unit can receive and process the sensor data for display when required. It is preferred for open water swimming to combine a GPS sensor with triaxial accelerometers to provide the essential velocity, distance, direction and stroke information. The accelerometers in combination with the processor clock provide information relating to stroke type, number of strokes per lap, lap times, turn efficiency and velocity off the wall. The velocity measurements from accelerometers tend to drift and the GPS signals are used to correct the velocity measurements in open water. The GPS can also be used to provide directional and distance information in open water as well as the same information in running and cycling. For tri athletes the accelerometers can also provide stride information and for the bicycle leg cadence information on the number of pedal revolutions. The central processor can also be in communication with a physiological sensor such as a heart rate monitor mounted on the athletes chest.
The quality of the display is an important issue particularly for swimmers. In a preferred aspect the wrist mounted display provides different coloured screens for preprogrammed functions. For example if the athlete is attempting to maintain a heart rate within a certain band a first colour indicates that the rate is within the band and a second colour indicates that it is too low and a third colour that it is too high. In open water swimming one colour may indicate that the swimmer is on course while a second colour may indicate that the swimmer needs to bear to the right and the third colour that the swimmer needs to bear to the left. In the pool tumble turns provide an opportunity for the swimmer to view a wrist display. As accelerometers allow the turn to be identified and also indicate the conclusion of a lap, the processor may be programmed to display on the wrist basic information such the lap number and the last lap time. Other information that could be displayed are heart rate and the number of strokes. Such a display is preferably for a short time as the swimmer comes off the wall. The display may be brightly lit for this period and be relatively large. The display can also be oriented for easy viewing on the wrist when the arms are extended in front of the swimmer which is the usual orientation coming out of a turn.

DETAILED DESCRIPTION OF THE INVENTION

Particular embodiments of the invention will be described.
FIG. 1 is a schematic layout of a data logger used for a rower and a rowing shell;
FIG. 2 shows the software output flow diagram for a rowing data logger;
FIG. 3 is a graphical illustration of stroke determined by using GPS data;
FIG. 4 illustrates the display for a computer screen;
FIG. 5 illustrates the deviation between code and carrier derived velocity measurements;
FIG. 6 is a video frame and triaxial accelerometer readings for a swimmer;
FIG. 7 is the core circuit diagram for monitor mounted on a swimmer;
FIG. 8 is the radio transceiver circuit for the monitor of FIG. 7;
FIG. 9 is the sensor circuits for the monitor of FIG. 7.
Recent developments in micro-electromechanical systems (MEMS) technology have opened new avenues for the use of high precision lightweight accelerometers and gyroscopes for new and challenging sports applications (eg. characterize rate and length of rowing stroke and stride). MEMS integrate both electrical and mechanical components on a single chip through extensive research into integrated circuit processing technologies. As MEMS accelerometers originated from monitoring vehicle safety and electronic stabilization, they only provided very low accuracy measurements. However, as micromechanical devices are inherently smaller, lighter, and usually more precise than their macroscopic counterparts, more and more reliable sensors are becoming available. Accelerometers measure linear acceleration and gyroscopes measure angular acceleration (pitch, yaw and roll).
Most accelerometers are used concurrently with gyroscopes to form an inertial navigation or “dead reckoning” system. That is where the deviation from position of a known reference (or starting point) is determined by integration of acceleration in each axis over time.
Inertial sensors errors include initial system heading errors, accelerometer scale factor and bias errors. These drifts and biases inherent in the inertial sensors will cause a misalignment of the platform and errors in the sensed accelerations, which subsequently results in errors in computed velocities and positions.
The advent of the advanced global navigation satellite systems (GNSS), GPS in particular, has revolutionized conventional precise positioning techniques. GPS has been made more amenable to a wide range of applications through the evolution of rapid static and kinematic methods, and now even more so with the advent of the On-The-Fly (OTF) technique and most recently network-based RTK techniques such as the Trimble virtual reference station system and Geo++ surface correction parameter method. Real-time Kinematic (RTK) or single epoch positioning allows for the determination of the integer ambiguities in real-time. It is therefore not necessary to carry out any static initialization before performing the survey. Due to the small wavelengths of the carrier phase frequencies (ëL1_iÖ19 cm and ëL2_iÖ24 cm), the determination of position within a specific cycle to a millimetre level by utilizing differential carrier phase measurements (i.e. differential techniques) is possible. Most systems statistically determine the most likely solution for the position of the roving receiver. Virtually, all carrier phase processing algorithms that utilize an OTF technique, rely on the double difference carrier phase observables as the primary measurement. A search box is determined within which the position must lie. All possible solutions are then assessed and the statistically most-likely candidate is selected. This procedure is extremely computing intensive, particularly with a large number of satellites.
Regardless of whether the system is for real-time or post-mission use, the algorithm is generally treated the same. Clearly, with real-time implementations, data outages, unfavourable observation environments, multipath and cycle slips can severely limit the performance of the system. The time for ambiguity resolution can range from a few seconds to several minutes depending on some of the following considerations:

- Use of L1 versus L1-L2 (widelane, L2_iÖ86 cm) observable
- Distance between reference and roaming receivers
- Number and geometry of satellites
- Ambiguity search method used and differential atmospheric conditions
- Quality of the received signal (multipath effects, code and carrier phase noise etc.)

Precise detection and removal of cycle slips is essential for the successful use of the OTF kinematic GPS technique. Various cycle slip detection techniques have been developed in the past decade. Included are double and triple differencing techniques, comparing the difference between adjacent carrier phase and code values (range residual), comparing the adjacent four observables equation, comparing adjacent ionospheric residual, the least-squares ambiguity decorrelation adjustment, carrier phase curve fitting, using redundant satellites and using the raw Doppler values. These methods typically assume a known stochastic behaviour for un-modeled errors (e.g. noise, multipath, differential atmospheric effects), which if present, will adversely affect the performance of the algorithm. None of these techniques can “cure all” kinematic positioning problems. Sometimes a cycle slip may be detected, but not accurately corrected for. Such instances include a loss of lock, large multipath effects and lower signal-to-noise ratio. This necessitates the combination of two or more of these techniques for a more robust solution.
FIG. 1 illustrates the basic components of a system to monitor boat speed and an oarsman's heart rate.
The accelerometer provides a PWM output where the duty cycle is related to the acceleration. On the rising edge and falling edge of the PWM output, a timer value is captured and used to calculate the accelerometers duty cycle. The firmware also includes an algorithm to adjust for jitter in the PWM period, and for a small amount of drift. A more detailed algorithm that compensates for temperature drift over time has been looked at, and will be implemented at a later date.
The impeller pickup uses a Melexis MLX90215 Hall Effect sensor to detect the rotations of the NK impeller. The MLX90215 is programmed with a sensitivity of 100 mV/mT. Output from the sensor is amplified by 100 to increase the signal amplitude to a usable range. This signal is then sampled using an A/D at 1200 Hz and processed using DSP techniques within the firmware to calculate rotations. Instead of using an impeller to detect boat speed a water flow sensor may be used. One preferred sensor is a micro PCB or silicon based micro fluid flow sensor that uses a heater in combination with a heat sensor that measures the change in temperature of fluid flowing past the heater and sensor to determine the fluid flow rate which in this case is the water flowing past a fixed point on the boat hull. This can then be used to measure boat speed.
For competition and race profile analysis it is preferred not to use impellers or water flow sensors but rely on GPS and accelerometers.
The display device is a handheld Compaq iPAQ™ computer programmed to present the data in a form that is useful to a coach or rower.
It is preferred that the device have data logging and IrDA transfer capabilities which makes data storage on the unit of slightly less importance. However storing data on the unit makes sense as the raw data can be streamed into the device and the greater processing power of the unit chip allows for flexible software and display development.
The microprocessor is a Hitachi HD64F3672FP which stems from the H8/300H family. Its main features are:

- eight 32-bit registers OR sixteen 16-bit or sixteen 8-bit
- Serial communication Interface (SCI)
- 10-bit ADC (4 channels)
- 2 k bytes of RAM

The accelerometer unit is powered from a 9 Volt battery, which is regulated down to 5 volts internally. The dimensions of the accelerometer unit are 25 mm×30 mm×9 mm (smaller that the average matchbox). The cover needs to be splash proof but importantly the on/off buttons and start/stop buttons etc must be able to be accessed even when the rowers are wearing gloves.
All the chips that have been selected are amongst the smallest available in their range, the Hitachi HD64F3672FP measures on 12 mm×12 mm, this incorporates a 64 pin architecture and the ADXL202 measuring only 5 mm×5 mm.
FIG. 2 illustrates the output flow from the various sensors namely impeller, heart rate monitor, clock, GPS sensor and 3 D accelerometer. Stroke rate and stroke drive to recovery ratio are most conveniently derived from the accelerometer data while intra stroke velocity, distance per stroke and velocity per stroke are derived from the accelerometer, GPS and time clock data.
The data for 1 block (by 3 or 4 channels) will be packaged and transmitted in a single frame. The sampling time for a frame (1 block at 150 samples/sec) will be equivalent to 6.6 ms. This data will be combined with block and channel information.

A total of eight bytes is required to transmit one block of data this includes the header, two 16-bit channels, Impeller Rotation count and Heart Rate count. The Heart Rate count is only transmitted once a second, or one in every 150 frames. Heart rate is an output indicating the millisecond value from the previous beat or the millisecond of the beat that occurred during that packet of information. This is used to calculate instantaneous HR on a beat to beat basis. Alternately the number of beats in 15 secs is totalled and then multiplied by 4 to get the HR. The algorithm then runs on a 5 sec rolling average to smooth the data. Given that maximum HR will never exceed 250 bpm this means that at most a beat will occur every 240 ms which is approximately 1 pulse every 2 packets of information. Table 1 shows a block of data excluding the framing and network information data.

TABLE 1


Byte	1	Frame header(xEE)
	2	Number of Blocks(4 bits)
		Number of channels (4 bits)
	3	ACC “Y” bits 1-8
	4	ACC “Y” bits 9-16
	5	ACC “X” bits 1-8
	6	ACC “X” bits 9-16
	7	Impeller rotation count (8 bits)
	8	Heart rate count (8 bits)

Table 2 illustrates an example of the bit stream for 2 frames. The first frame containing two 16-bit channels and Impeller Rotation count, and the second frame containing two 16-bit channels, Impeller Rotation count and Heart Rate count

TABLE 2


Data Stream	Meaning

0xEE	Header Byte
0x13	One Block, eg. 3Channels
0xA9	Acc Y Lower Byte
0xEA	Acc Y Upper Byte
0x46	Acc X Lower Byte
0xC9	Acc X Upper Byte
0x01	Impeller Rotation Count
0xEE	Header Byte
0x14	One Block, eg. 4Channels
0xA9	Acc Y Lower Byte
0xEA	Acc Y Upper Byte
0x46	Acc X Lower Byte
0xC9	Acc X Upper Byte
0x01	Impeller Rotation Count
0x02	Heart Rate Count

A single unit may be used for each crew member or the heart rate lines for each crew member can be included with the accelerometer and speed data to provide a composite set of data. In a multi crew boat each crew member has a receiver within 2 feet that picks up the heart rate signal from the polar heart rate monitor strapped to each crew member. Each heart rate monitor transmits a uniquely coded signal that is assigned to each crew member the boat data logger receives the heart rate signals for all crew members by cable from the heart rate receivers
A GPS unit may be integrated with the data logger system. This could comprise two units, basic unit plus a second unit for GPS. The units would share the same serial line and communicate using a network protocol. Alternatively the GPS unit could be connected to the basic unit and additional firmware code added to receive and retransmit data.
Inertial navigation systems (INS) may be used to cover the information gaps of the GPS outages. When the INS approach is used in rowing, the required sensors need to be small, lightweight, unobtrusive and inexpensive. These requirements can be met when the sensors are manufactured with MEMS technology. However, due to inherent biases and drift errors of accelerometers and gyroscopes, the accuracy of the current state-of-the-art MEMS sensors must be accounted for in high precision rowing tracking. The basic procedure in INS positioning systems is to process the inertial sensor data. The double integration of acceleration measurements, cannot be applied due to the lower accuracy of MEMS sensors. This is because in the double integration, errors accumulate quickly, which soon result in velocity errors comparable to typical rowing speeds. However, the advantages of the INS system include its low cost and high output rate of the movement information.
The high precision GPS system can provide high precision velocity and acceleration information (acceleration is the first derivative of velocity and second derivative of displacement). However the GPS system is normally bulky, expensive and provides a low output rate and high power consumption. To solve these problems, an integrated system takes advantage of both low-cost GPS and MEMS sensors to provide high performance capabilities. MEMS sensors are used to provide precise, high rate (say 200 Hz), low cost, low volume, low power, rugged, and reliable geo-positioning while low-cost GPS is used for high frequency system calibration (say 5-20 Hz) a lower frequency (1 Hz) is preferred for calibrating the inertial sensors to conserve battery power. It combines measurements from a GPS OEM board and subsequently GPS chip with inertial measurement units from a combination of three MEMS gyroscopes and accelerometers (say Analog Devices).
A 1 Hz GPS receiver is the minimum frequency that is practical and ideally a 2-5 Hz system is preferred. With a 1 Hz receiver accurate velocity and distance measurements can be obtained but sampling the accelerometer data is needed to obtain stroke rate and intra-stroke characteristics. The accelerometer data could be integrated to get intra-stroke velocity but drift would need to be checked every second using the output from the GPS receiver.
The carrier smoothing procedure will be used to improve the accuracy of the low-cost GPS pseudo range measurements. Carrier phase smoothing is a process that the absolute but noisy pseudo range measurements are combined with the accurate but ambiguous carrier phase measurements to obtain a good solution without the noise inherent in pseudo range tracking through a weighted averaging process. A Kalman filtering system will be designed to integrate the two system measurements.
FIG. 3 presents the stroke signals captured using geodetic type GPS receivers and post-processing with the kinematic differential GPS technique. It is demonstrated that the signals captured provide a clear picture of the rowing stroke phases as described above. In this particular stroke, the graph indicates that the rower has problem in harmonizing his stroke cycle by using too much time in the catch instead of the driver.
The software can display the derived information on a computer screen and combine it with video data of the same event as illustrated in FIG. 4. The screen may display time and distance information as well as velocity and stroke rate and can also display the graphical signals derived from accelerometer and GPS signals.
To evaluate the accuracy of the GPS carrier phase receiver, two GPS receivers were mounted on the same rowing boat simultaneously. The base station is located on the bank of a river which is about 1˜2 km away from the course of the boat trial. The baseline solutions from each of the rowing antennas were processed independently from the base station using the PPK technique. The independent baseline length between the two roving receivers was then calculated and compared with the result measured using a surveying tape. This baseline length is considered as a “ground truth” (3.57 m in our case).
RTK GPS has been proved to be able to provide high precision positioning in river environment. However, there are a number of factors that need to be taken into consideration:

- Multipath effects: The antenna being positioned near the water surface could potentially be prone to large multipath error. This effect can be up to 5 cm for carrier and 5 m for code measurements respectively.
- Signal obstruction/satellite visibility: The GPS antenna is installed in a constricted space in a racing boat, it is therefore unavoidable that the movement of the athlete will block the GPS signals at some time to an elevation angle of approximately 70 degrees. This may potentially cause severe signal obstruction problems and loss of GPS solutions.
- Obtrusion: Ideally the presence of any instrument should not cause direct visual or physical impact on the athlete, therefore, the size and height of the antenna is a primary consideration.

The “fixed baseline length” and external check methods are used. Reliable mounting of the GPS receiver is required. If we assume that the accuracy of the position to one GPS rover is the same as to the other, then, from the simple (Least Squares Adjustment) error propagation law, the accuracy of the position of the kinematic GPS measurement (for a single baseline) can be estimated as 0.0027 m (0.0038 m/sqrt(2)). A few millimetre accuracy of the river height was achieved in a three (consecutive) day trial. Given the closeness of the antenna and the reflective nature of the water surface, the performance of the PPK GPS presents consistent results.
The velocity determined from the GPS position and time information uses the following first-order central difference procedure. $Velocity (υ_{T}) = \frac{P (T + Δ T) - P (T - Δ T)}{2 Δ T} = \frac{Δ P}{2 Δ T}$
where ν_Tis the velocity of the boat (at time T) determined from PPK GPS solution, ΔP=P(T+ΔT)−P(T−ΔT) is the plane distance travelled between time T₁and T₂and ΔT=T₂−T₁. $Δ P = \sqrt{{(N_{2} - N_{1})}^{2} + {(E_{2} - E_{1})}^{2}},$
where E and N are the Easting and Northing coordinates of the GPS units. The subscripts “1” and “2” indicate that position derived from unit 2 and unit 1 respectively. The accuracy of the velocity (σ_ν), can then be roughly estimated through the following formula (using the error propagation law): $σ_{υ} = \frac{1}{\sqrt{2} Δ T} σ_{P} = \frac{1}{\sqrt{2} \times 0.1} \cdot 0.0027 \approx 0.02 m / s$
Where σ_Pis the positional accuracy and σ_P=0.0027 m as determined previously.
FIG. 5 shows the differences in velocity determined simultaneously from the code and the carrier measurements. Assuming the carrier velocity to be accurate (ie ground truth), the code derived velocity has an average accuracy in the order of ˜0.03 m/s. The results confirm that the accuracy of 0.1 m/sec claimed by the manufacturer is correct for more than 95% of observations.
The data logger assembly is fitted to a rowing shell in a stable location with a relatively clear view of the sky. Relative motion of the athlete or boat is measured using three dimensional accelerometer at 100 hHz and position and velocity using GPS at 10 Hz. The device supplies timing information with the measured signals using an internal crystal corrected clock and a GPS derived 1 Hz pulse. The timing is accurate to 0.1 sec per hour. An internal heat rate monitor pickup receives pulses from a coded polar heart rate monitor/transmitter and stores these with a resolution of 1 beat a minute within a range of 0 to 250 beats/minute updated at 1 Hz. The device is powered by a battery sealed into the unit and is rechargeable via an RS232 port. Recording battery life is 6 hours and 1 month in sleep mode. The single universal port allows recharging, connecting an RF module, connecting an external GPS antenna, connecting the external heart rate receiver and to connect a serial cable to send data to the hand held computer device. The device can be fitted into a flexible package of a size approximately 100 mm×70 mm×50 mm and weighs less than 250 g and is buoyant and water resistant. The package is coloured to reduce heating from incident sunlight.
The device can be adapted to detect strokes and turns in swimming as shown in FIG. 6. Analysis of the signals from the 3 axes of the accelerometer allows coaches to derive information as detailed as stroke formation and turn efficiency.
FIGS. 7 to 9 illustrate the circuitry used in further embodiment of the invention.
FIG. 7 shows the core circuitry centred on the micro controller 20. The micro controller is preferably an 8 bit Atmel AT mega 128 micro controller. The micro controller can be programmed and can store data and is provided with a 256 megabyte flash memory 27. The USB port 22 is preferably a Silicon Technologies USB to UART data transfer CP 2101 and allows data to be down loaded to a personal computer for further analysis and storage and also allow the battery to be charged by way of the battery charger 31 which in turn is connected to the power supply 32. The microcontroller functions are actuated by the tactile switches 23 which allows the user to navigate through the device menu. The microcontroller displays outputs on the LCD display 35 and also provides a backlight display 36. As shown in FIG. 8 the monitor includes a 2.4 GHz transmitter and receiver 40 so that data can be transmitted and received. The transmitter and receiver 40 is preferably a GFSK transceiver nRF2401 sold by Nordic Semiconductor. The output power and frequency channels are programmable using a 3 wire serial interface. The GPS unit is an iTRAX 03 by Fastrax with 12 channels and an update rate below 5 Hz with a 1 Hz default rate.
The sensor circuits shown in FIG. 9 are the core components of the real time clock 41 the three axis accelerometer 43. A single external transistor may be used to lower the scale factor and an external capacitor is used to set the bandwidth.
The micro controller in the swimming monitor is programmed with a set of algorithms to process the raw data from the sensors. The algorithms filter the raw data with a low pass filter. The aim is to use slower changing orientation information from the accelerometers rather than quickly changing real accelerations. The algorithm looks for peaks and troughs on each filtered sensor trace. Strokes are defined as combinations of peaks and troughs—each stroke type has a specific combination with a specific set of rules. Once locked on to a particular stroke type then look first for that stroke type next. Initially looks for freestyle first.
Freestyle

- ignores small peaks/troughs
- requires up/down accelerometer>0
- requires a sideways accelerometer peak followed by a trough—each peak/trough is a stroke
- looks for several strokes in a row to lock on

Backstroke

- ignores small peaks/troughs
- requires up/down accelerometer to be less than 0
- requires a sideways accelerometer peak followed by a trough—each peak/trough is a stroke
- looks for several strokes in a row to lock on

Butterfly

- consists of two peaks—one higher than the other
- looks for several fwd/back peaks in a row to lock on—first must be high, next low, next high etc
- peaks must be spaced appropriately
- high peaks should be equally spaced, low peaks likewise
- high peaks should be equal magnitude, low peaks likewise
- highest up/down acc peak in the area must be large enough
- lowest up/down acc peak in area must be significantly less than highest

Breaststroke

- uses troughs in fwd/back acceleration
- two quick troughs and a gap
- looks for several troughs to lock on
- sufficient trough spacing
- time between troughs ½ and ¾ should be close
- time between toughs ⅔ and ⅘ should be close
- up acc must be >0

There is a fourth type of stroke which is the dolphin kick.
Starts/turns/Finishes
A state variable keeps track of the current lap state. There are 3 possible states:

- Waiting for a start
- Progress during the lap
- Possible end of lap

Waiting for Start
When a stroke is detected in this state look backwards for the start. Since stroke detection requires several strokes in a row (depending on the stroke type) then we are likely to be a fair way down the pool at this stage, particularly after a block start and a few dolphin kicks (these are ignored for start purposes—there has to be several of the regular stroke types in a row before checking for a start).

- After the first stroke look at rate-of-change peaks. If there is only one, or the highest is large enough, then we have a start at the highest point.
- If the above didn't succeed then look for a large swing in z (up/down acc) in the time before the first stroke—this is defined as a peak >0 g preceded by a trough <0 g and with sufficient difference between the two. The start is then the low fwd/back trough which is close to the highest rate-of-change in the region.
- If neither of the above get a result then the start is a fixed time before the start of the first stroke.

State then changes to . . .
Progress During the Lap
After being in this state for sufficient time, watch for turns or end of lap

- First look for low fwd/back acc readings either side of the end of the last stroke. If the lowest trough before the end of the last stroke is sufficiently greater than the lowest trough after then change state to “Possible end of lap”
- If above wasn't successful then look for a large vertical accelerometer swing. Again look either side of the end of the last stroke. This time search for a z-axis high to low change to change state to “Possible end of lap”

Within the time it is also possible to change state but only if enough time has elapsed with no sign of a new stroke.
End of Lap Detection
At next stroke look back from the start of the last stroke for the end of lap:

- A large swing in z (as above)
- Or a large rate-of-change (as above)
- Or the lowest fwd/back acc reading
- Or a point a fixed time period back

Also look for the finish of a set of laps.
This is done by looking for the first point with “zero” rate-of-change which is defined as a short period with all rate-of-change results (ie for every point) sufficiently low. If this is found and there are no strokes for a while then end of set is assumed to be at the beginning of the “zero” rate-of-change period.
A display is mounted in a water proof enclosure in a visible location so that the athlete can view summary information such as stroke rate distance and heart rate. An easily accessible button on the display unit starts the data recording. As soon as the device is switched on recording begins. The coach may take the device after the event and load the data into a personal computer to view the data graphically or combine it synchronously with video footage.
The device may be attached near the small of the back. An extension GPS aerial runs from the device to the shoulders, but mounting on the wrist or head is also possible. A separate GPS unit may be mounted on the head to improve reception and the microcontroller, accelerometers and display may be mounted on the wrist or arm.
The RF module enables the real time data to be transmitted to the Coach's wireless enabled PC via a blue tooth connection. Alternatively the data may simply be uploaded after the event.
The advantages of the swim monitor of this invention are:

- The device can give feedback both in real-time and post-training.
- The real-time feedback to the swimmer may be via a variety of methods:
  - Aural via an ear-piece
  - Visual via a ‘heads-up’ display on the goggle
  - Visual via an LCD panel on the device
  - Visual via a remote display panel either in the pool or adjacent to it.
- If the preferred real-time feedback is used, some additional circuitry incorporating an FM receiver may be used to allow a coach to talk to the swimmer via the device.
- The real-time feedback would likely be delivered at the start of a new lap and would give key results about the previous lap. The results may include:
  - Average velocity
  - Number of strokes
  - Number of laps completed
- A further enhancement of the system would be to add an MP3 player or FM receiver to the device so the swimmer may be entertained.
- The device may also include pulse counters taking advantage of the temple mounting for deriving heart rate
- For tri-athletes the device may include accelerometers or GPS units to derive speed and distance and stride length data for the land based activities
- The post-training display may be on a standard PC. It would show summary graphs such as ‘velocity vs time’ and ‘stroke-rate vs time’ for the entire session. It would also be able to calculate bio-metric efficiencies such as distance per stoke. The user is able to zoom into a section of the graph to obtain information about each stroke, enabling the swimmer to gain information about how bio-metric improvements may be made.

Those skilled in the art will realize that the invention may be implemented in a variety of embodiments. A variety of sensors may also be used to gather data applicable to the event. It will also be appreciated that the logger unit is small and adaptable enough to be fitted to any athlete or sporting equipment where accelerometer data provides useful performance information for coaches and athletes.

Claims

1. A data acquisition system for use in swimming events which incorporates

a) a global position sensor to derive three dimensional positioning data relative to time elapse

b) at least one accelerometer to derive acceleration and velocity data in three dimensions

c) a microcontroller with a clock to interrogate the global position sensor and to collect the accelerometer data

d) a power supply

e) communication means for transmission of global position and accelerometer data from the microcontroller to a computer device

f) the computer device being programmed to use the global position and accelerometer data to provide accurate and continuous output of parameters such as velocity acceleration and distance traveled.

2. A data acquisition system as claimed in claim 1 in which velocity is derived from the global position sensor and the accelerometer data is sampled to obtain movement characteristics in swimming

3. A data acquisition system as claimed in claim 1 wherein the accelerometer data is integrated to derive velocity related movement characteristics and drift is be checked every second using the output from the global position sensor.

4. A data acquisition system as claimed in claim 1 wherein an inertial navigation system based on the accelerometer data is used to determine position when the GPS system is unable to receive data.

5. A data acquisition system as claimed in claim 1 which includes a display screen.

6. A data acquisition system as claimed in claim 5 in which the global position sensor is located in a separate unit to the micro controller and the display.

7. A data acquisition system as claimed in claim 6 in which the global position sensor is adapted to be mounted on the swimmers head and the display unit is adapted to be mounted on the swimmers wrist

8. A data acquisition system as claimed in claim 1 which also includes a physiological sensor.

9. A data acquisition system as claimed in claim 6 in the physiological sensor is a heart rate monitor.