Optimized obstacle avoidance trajectory generation for a reconfigurable staircase climbing wheelchair

摘要

This paper describes the mechanical devices, the control scheme and the trajectory generation of which a new wheelchair prototype capable of climbing staircases is formed. The key feature of the mechanical design is the use of two decoupled mechanisms in each axle, one to negotiate steps, and the other to position the axle with regard to the chair in order to accommodate the overall slope. This design simplifies the control task substantially. Kinematic models are necessary to describe the behavior of the system and to control the actuated degrees of freedom of the wheelchair in order to ensure the passenger's comfort. The choice of a good control scheme based on a local and a global trajectory planner simplifies control, decreases power consumption, reduces the time invested in traversing the obstacles and maintains passenger comfort throughout all movements. The paper presented here is the natural continuation of a previous work presented in [R. Morales, A. Gonzalez, V. Feliu, P. Pintado, Environment adaptation of a new Staircase climbing wheelchair, Autonomous Robots 23 (2007) 275-292]. After studying the time outs in the staircase climbing/descent process due to configuration changes, we started to increase the capabilities of the trajectory planner in order to reduce the time invested in traversing obstacles. The optimization algorithm is only used in the period of time in which configuration changes are being produced. More specifically, we have used the special properties of the mechanical configuration, the kinematic model and the trajectory planner to develop an improvement in the trajectory planning based on complex notation. The new optimized algorithm solves a nonlinear problem in order to discover an auxiliary center of mass route which is free of obstacles, through the work environment of the wheelchair prototype. Additional properties of the new optimization algorithm are: (a) the resulting analytical expressions are closed (iterative calculation is not necessary); (b) it is easy to implement in the real prototype and (c) it can be executed in real time. Experimental results are reported which show the behavior of the prototype as it climbs a staircase both when using the original trajectory planner and when using the new obstacle avoidance optimization algorithm explained in this paper. The results obtained illustrate a high percentage of time reduction and the maintenance of comfort levels. However, the control prototype becomes more complicated, the power consumption is increased and the comfort level is slightly lower.