Patent Application
20170144047
None
Not Applicable
Not Applicable
Not currently aware of relevant prior disclosures.
1. Field of the Invention
The general field of the invention is rowing technologies. The invention relates to systems and methods for tracking the performance of individual athletes and boat components in close detail so as to better ascertain their effectiveness and make needed corrections.
2. Background
The sport of rowing (and sculling—not differentiated for the purpose of this application) involves a significant number of mechanical and physiological factors that interact in a complex way. Individual rowing performance is affected by the biomechanics (how well you handle the oars, how well you handle the sliding seat, how well you position yourself to transfer power between the oar and the shell, etc.) and your physiologic ability to systematically transfer energy from the oar to the shell. When multiple people are in the boat, additional complexities arise in orchestrating the actions of all the rowers. Coaching plays a critical role in improving rowing proficiency and success.
Rowing coaching is traditionally done via a boat riding alongside the rowing shell with the coach shouting instructions to the rowers and/or coxswain. This form of coaching inherently creates numerous limitations as the coach can only work with what he/she is witnessing, analyzing, and instructing. Furthermore, the multiplicity of rowing factors and their complex interactions makes it very difficult if not impossible to quantitatively decompose rowing success (or failure) into its constitutive elements. Thus, rowing coaching (including self-coaching) largely has become an art developed through years of experience. In engineering terms, the “feedback loop” is neither precise nor tight.
This invention describes a system that uses sensors, communication systems, networks, standards, models, computing devices, and machine learning techniques to build deep insights into rowing cause/effect relationships and to create a tight feedback coaching loop. This system can work by itself or in concert with the traditional coach curating and augmenting the feedback. The system works in real-time during rowing workouts and races and as well as a post-workout assessment and planning tool. A particularly novel effect of this system is the ability to understand the sensitivities of rowing performance to proposed changes and thus create a prioritized set of coaching instructions. Additionally, the system makes it possible to tease out and distribute contributions of the shell performance to individuals within the boat.
The system involves location, Orientation and Motion sensors placed on the shell of the boat, the oars, and the rower seats. The sensor systems are coupled as necessary with other systems and data sources needed to deliver their functionality, e.g., GPS satellites, world/atomic clocks, network time standards, network-based localization techniques, environmental data (weather, tide, etc.), physiologic sensors on the rowers (heart rate, respiration rate, perspiration, body temperature, etc.) and other contextually relevant information. The Sensor Packages each have processing and communication functionality that allows the sensor data to be transferred and analyzed as needed within personal area networks (PAN; one for each rower), local area networks (LAN; the shell, rowers, and coaching boat), and wide area networks (WAN; providing additional analysis resources). See
The system uses the sensor information with other contextually relevant data and employs standard machine learning techniques to construct a model of the rower/shell system that is specific to the training or racing configuration. This model is iteratively refined as additional “training data set” information is accumulated throughout a rowing event. The model can then be used to conduct sensitivity analysis of boat performance as a function of perturbations about each rower action. The effect of each perturbation can then be used to quantitatively predict the resulting change in boat response and thus create a prioritization of coaching instructions based on its predicted impact on shell performance. Each set of data (vector) on oar and seat Motions for each rower is compared to one (or more) standard of rowing mechanics and deviations are identified for potential corrective instruction and scored to measure training progress. A generic characterization of the model development methodology is depicted in
The system can utilize several different feedback methods and channels based on the chosen coaching criteria and training/racing plan. Individual rower analysis can be conducted at the personal area network level with standard machine learning classification techniques to identify and correct identified deviations of oar and seat management from assigned standards, e.g., “rushing the slide”, “hands away too slow”, “oar too deep” could be communicated to an individual rower. The individual rower data and shell performance data can then be assembled within the local area network to build insights into the individual and collective effects of what is going on with the shell. For example, the system might identify that seat 3 is consistently late in their catch timing relative to seat 4 and thus suggest an appropriate corrective action. Lastly, all of these data can be streamed through a wide-area network to resources that store and process this data for a number of purposes, including but not limited to more complex sensitivity analyses, simulations, grading of rower performance, and additional refinements of models, assigned rowing standards, and training/racing plans.
The system could be configured to work in an automated feedback fashion in accordance with a defined criteria. For example, feedback could be blocked for certain instructions during certain workout time periods or the feedback could be limited to instructions that would be expected to achieve performance results that exceed a certain threshold. The system can be made to work in concert with a coxswain and/or coach as a higher level authority that is approving or blocking instruction based on more global information, e.g., a grander training plan, individual athlete psychologies, conditioning priorities, etc. The system may utilize several different communication channels. One channel may issue discreet individual instructions to each rower through Any number of channels, e.g., an audio earpiece that only the individual rower can hear, a tactile feedback system that only the rower can feel, a visual channel that only the individual rower can see. Another channel may involve a more general broadcast to the boat through audio, visual, or tactile systems either through a speaker, coxswain, or coach instruction.
The system can synthesize the results of Any portion of the workout into Any number of scoring systems relative to a chosen standard or metric. For example, an oar or seat control metric that scores locations, velocities, and accelerations against a known standard could be derived from the workout and published. The scores could be compared to trends for a specific rower or a group of rowers to measure their progress against a prescribed learning curve. With the use of system that can measure individual rower actions, shell performance, and a model of the relationship between these data, it would be possible to predict the energetic (work) contributions that rowers individually make on a shell. Thus, it will be possible to quantitatively assign and rank rowers based on their individual “boat moving” index.
One embodiment of the invention involves the use of smartphones and a resident application to control the use of the integrated smartphone sensors, network and communication systems, and processing functionality. Modern smartphones universally possess precision clocks, PAN/LAN/WAN network communication functionality, GPS functionality, and inertial measurement units that measure linear accelerations, angular velocities, and magnetic fields in all 3 degrees of freedom. A smartphone could be securely attached to an oar though Any number of fixation techniques (armband, straps, Velcro, wraps, etc.) and be used to measure oar location, Orientation, and Motions through the inertial measurement unit sensors. Of particular interest in the rowing environment will be the location, rotation, and rotation rates of the oar blade throughout the stroke cycle (catch, drive, release, recovery). Another smartphone could be attached to a rowing seat or waist of a rower to measure the linear Motions of the seat along the seat track and used to determine the so-called slide control and that the rower is effecting. Synchronizing the oar and seat/waist information within the PAN makes it possible to analyze the sequencing of the rower's hands and legs. Within this personal area neatwork, there may be additional actuators/devices to deliver feedback, e.g., watches with haptic engines, ear pieces with speakers, or LEDs for communicating information. Another smartphone can be attached to the boat through Any number of fixation techniques, e.g., suction cups, straps, Velcro, etc. and the GPS functionality and IMUs could be used for high precision measures of boat velocity, Orientation, and dynamics, e.g., impulse/accelerations associated with each stroke of the boat.
Another embodiment of the PAN elements in the system could involve simple peripheral devices like activity trackers in place of a smartphones. For example, a smartphone could be affixed to the oar as described in the previous embodiment and be paired in the PAN (e.g., via Bluetooth) with a simple activity tracker attached to the rowing seat. Together, the smartphone and activity tracker would constitute a complete sensor, communication, and processing system for an individual rower.
An embodiment of the model derivation methods involve machine learning algorithms like time-series analysis, neural networks, Markov models, Bayesian networks, and ensemble techniques. It is anticipated that in addition to real time measures of rower mechanics and shell performance, there are contextually relevant information intrinsic to each rower that will prove to be important, e.g., individual power output measured on a rowing ergometer, real-time measures of heart rate, historical measures of the aforementioned “boat moving” index, etc.
An embodiment of the coaching analysis logic could involve techniques that learn to classify data in the language of rowing by using “training data” classified by an expert. For example, a machine could be taught to recognize and classify ‘washing out’, characterized by pulling the oar out of the water too early, by using data collected on oar handle position and velocities classified and validated through video by a coach as ‘washing out’ or ‘not washing out’.
An embodiment of the coaching instruction assessment could involve the check against a set of predefined criteria, e.g., 1) communication of an identified problem that reduces boat speed more than 1% on 3 consecutive strokes or 2) a system of cued recommendations in the coach or coxswain system that lists actions prioritized by their effect. The coach or coxswain could select the instructions and the means of communication on a smartphone or tablet with in the LAN.
One embodiment of a grading system could involve the use of a function that scores performance based on penalties for deviations from an established norm or model standard(s). The oar and seat/waist Motions recorded for a rower over a chosen series of strokes could be compared to those of an idealized Motion or those of an elite rower standard with a grade assigned based on some desired bound, e.g., “oar position within 1 inch of the standard 81% of the time” or “oar position within 2” of standard 94% of the time”. These grades could further be compared to some demographic or experience-based norm, e.g., “scores are 74th percentile for rowers in your age group and 38th percentile for rowers with 10-15 years of experience”. Of particular interest to an athlete would be how these grades evolved over time, i.e., how one is progressing against a target learning curve.
One embodiment of the individualization of rowing performance could involve the simple aggregation of the estimated translational energies imparted on a boat over a set period of time or event. For example, the rowers in seats 3 and 4 may be responsible for 18% and 20% of the energies generated for shell speed in drill 1 but only 11% and 15% of the energies generated for shell speed in drill 2.
One embodiment of the peak performance elicitation is utilizing information on current or historical physiologic status (HR, lactic acid status, rowing efficiency/effectiveness, etc.), contextual or environmental conditions, and sensitivity analysis of the model to perturbations to maximize the probability of a desired result. It could be that tide or environmental conditions during a race will effectively make the distance longer or shorter than planned and thus justify real-time race replanning.
The invention consists of the following elements:
This application claims priority to an earlier-filed provisional application No. 62/258,368. Confirmation No. 2808. Filed Nov. 20, 2015.
| Number | Date | Country | |
|---|---|---|---|
| 62258368 | Nov 2015 | US |
