Up

Ambient Assisted Training

The AmI system behind this demonstrator has to optimise the training of a group of racing cyclists.


Szenario
up

There is group of 2 to 30 racing cyclists. Each cyclist has his own performance profile and training plan. Cyclists and bikes are equipped with a multiplicity of sensors (speed, pulse, pedal force, power, wind, etc.). The bikes are connected to each other and to a coach unit that executes the application, via an ad-hoc, wireless network. The task of the AmI system is to optimise training for the entire group. The training ride has to be planned and controlled in such a way that each cyclist meets his training plan as far as possible.

In an additional scenario, the AmI system also has to support the overall competition, that is, the race situation of the bicycle group as well. Using the same technology, the task now is to win a race; that is, a given point from A to B must be passed in minimal time by one member of the group. The bicycle group is controlled using the stored performance data as well as the actual sensor data, and the condition of the cyclists.

Of course, the training of an individual cyclist can also be supported with such an AmI system.

The advantages of an AmI system in a training situation will now be summarised using two examples, individual training and group training.


Individual Training

The AmI system gives an online optimisation for training stress based on an individual's performance profile and training plan. Here, the control values of "power" and "heart rate" will be adapted in order to process their dynamics. The values will be modified depending on the individual's physique and individual strength as well as environmental and track conditions (wind, track profile, etc). The system thereby enables online expertise and intervention by the coach during the training, as well as training documentation and analysis after training.

Two suggested scenarios could be as follows:

  • The heart rate of the cyclist is below a given threshold value for the given and obtained pedal power. The physical load is considered too low. The system increases the pedal power until the target heart rate is reached.
  • During an up-hill ride, the system determines an unusually high and steep rise in heart rate. The cyclist receives the recommendation to lower his speed or to change down a gear.


Group Training

This scenario also concerns the optimisation of the training load of each individual cyclist, however, the cyclist is riding in a group. The AmI system must consider the restrictions of the group, so that the optimisation of the overall performance of the group remains the focus. The formation and speed of the group as given by the system depends on:

  • track profile
  • position of the group on the track
  • ability and fatigue of the cyclists
  • environmental conditions (eg wind, weather, etc)
  • training plan

The AmI system is now used to monitor the training parameters of the individual cyclists to optimise the group's speed, position changes, and formation/structure.

For the group training, two possible scenarios are presented:

  • The workload of the leading cyclist who faces into the wind is subject to more intensive stress from which they can subsequently recover by cycling behind the leading cyclist. Thereby, if the heart rate drops slowly due to the fatigue, the system varies the order of the position changes in such a way that this cyclist obtains a longer break and a less fatigued cyclist faces the wind.
  • Two cyclists at the end of the formation can only remain as part of the group when cycling at a higher heart rate than specified. In order to avoid overload during the training, the two cyclists are separated from the group and form their own formation.


AmI-Relevance
up

Although the scenarios described above look like toy scenarios without any serious background, this is not true. Many trainers are very interested in having such a training aid. Whilst the whole system is far beyond state of the art, several aspects are already available for top teams, who can afford current, expensive solutions. Of broader interest would be an "AmI solution" for the whole scenario that is cheap and light weight, comprises unobtrusive hardware, and that controls and adapts to an actual situation. For that, the objectives have to be elaborated.

For our AmI research center, the main advantage of this scenario is that it covers many different research questions, which are typical for AmI:

  • Energy efficient with minimal sensors/hardware
  • Ad-hoc communication
  • Control engineering
  • Training methodology

Today, the tasks of the AmI system, which have been described, are done more or less optimally by a coach. The coach generally knows the quality of the cyclists. However, he does not know the exact, current situation regarding training or running. "However, the Aml system could further improve the situation dramatically," confirms Mr Mühlfriedel, head of the racing cycle department at the "Heinrich-Heine-Gymnasium", an elite sport school not far from our university.

In addition to their own research questions, the scenarios relate to other domains so that solutions for these scenarios will fit others, too. Examples for general AmI research questions include:

  • Human orientation
    First of all, our AmI system supports human activities. Its main objectives are to improve the physical condition and monitor the health of human beings doing sports. The system has to adapt to the cyclists' activities and condition therefore it requires situation- and location-based functions (the goal depends on location, physical condition, trainer, etc).
  • Human-Computer-Interface / Usability
    As described above, the user interface has to be situation- and role-based. A cyclist may have different roles (e.g. leader, trainer) and interests. Exceptions must be covered (e.g. health monitor, too much headwind). The dynamic integration of many different and new device types (e.g. glasses with a display or a bike helmet with microphone and speakers) requires the development of new device models with bespoke computer software.
  • Sensors
    Sensors should be simple and cheap. The sum of many simple and coupled sensors for various physical values should result in more information than we have today using current, expensive, stand-alone systems. In the scenario, sensors need to be small, lightweight, wearable, unobtrusive, and robust. It is possible that newly developed sensor types can improve the application.
  • Dynamic communication network
    A central aspect of the scenario is the development of a tailored, wireless, dynamic, ad-hoc network based on different technologies. Sub-networks of a hierarchical communication system (BAN, LAN, WAN) may disturb each other drastically. Communication channels will vary all the time, because cyclists vary their formation all the time, too. Multi-hop routing requires (sub)networks to be split and reconnected.
  • Complex control
    The AmI system has to control a dynamically changing group. The control objective depends on the situation, the actual track profile, and the physical condition of the cyclists. Changing communication paths results in varying latency times.
  • Dynamic system architecture
    The variable hardware platform addressed briefly above requests a dynamically changing system architecture. For instance, functions which are located on a powerful PDA mounted on the handle-bar during training (e.g. due to dependability reasons) must be moved to a remote computer during the race so that cyclists only use small, lightweight cycling computers.
  • Technology constraints
    Our scenario covers almost all technology constraints which are related to AmI. On the one hand, we need a dynamically changing hardware platform (e.g. totally different processors and user interfaces, plug & play of sensors) and software structure. On the other hand, we need low power (all sub-systems are powered by small batteries), restricted size (hardware devices including sensors are mounted on the bike or cyclist), and something light weight (for racing cyclists each gram counts).


Demonstrator
up

In mid-2004, our AmI Research Centre began to develop a demonstrator for the bicycle scenario. The demonstrator covers four racing bikes, which are equipped with different sensors and a wireless network (see below). The demonstrator is designed in such a way that in the medium-term, we can perform experiments with competitive cyclists in cooperation with the Heinrich-Heine-Gymnasiums.

outdoordemonstrator
Outdoor-Demonstrator


Furthermore, it is hoped, that the demonstrator can be used both outdoor ("outdoor" or "real life" demonstrator) and indoor ("indoor" or "VR" demonstrator). For the latter, the four bicycles, mounted on wheel stands or ergometers (ref. indoor cycling training) are attached to a simulator. This simulator, in combination with a screen and the braking force of the wheel stands, replaces the environment of the real-life demonstrator. By adapting the braking force for each bike individually, cyclists with totally different physiques can use the demonstrator together.

indoordemonstrator
Indoor-Demonstrator


indoordemonstratorscreen
Indoor-Demonstrator visualisation screenshot



Realization
up

We currently work on following tasks:

  • Advanced low power sensor technology with inductive power supply for measuring the force at the crank. Depending on the angle of the crank, the force can be determined separately for the right and left side.
  • Electronics for input and preprocessing of different sensor data.
  • Indoor simulator: Simulation of bicycle+cyclist and environment.
  • Cross-layer optimized protocol for communication between the bicycles via ZigBee.
  • Software platform which enables the dynamic integration of new devices (e.g. displays sensors). This platform is also used in the BelAmI demonstrator.

With the currently available prototypes we already performed first experiments, successfully:

  • Together with our Hungarian partner, we performed an experiment in the context of the BelAmI project regarding speech understanding under physical load.
  • An analysis of lactate and heart rate parameters within the context of the maximum lactate steady state: this experiment has been conducted by Ass.-Prof. Jaitner together with Dr. Bleckmann from the Medical Institute for Performance Diagnostics in sports.
  • Student project: Development of an AmI software system for a coachless individual and group training.


Visions
up

Finally, we want to list several visions in the context of the assisted training demonstrator:
Indoor demonstrator

  • Realistic training in the laboratory, e.g. usage in winter training or in the rehabilitation (various bicycle ergometer manufacturers work on this task, too).
  • Development of race strategies for the team time trial by simulation.
  • Race simulation.

Outdoor demonstrator

  • Monitoring the performance parameters of different disciplines during training and race.
  • On-line control of a group by the coach (in the following car) during training and race. The coach can be supported by suitable software assistants.
  • Temporal log-in of the coach who is e.g. at a fix position of the racing circuit.
  • Optimization of the team composition and alternations during team time trial.
up

Last update: 24.10.2007