Advanced control of robotic systems using low cost sensors

移动机器人必须执行涉及两个或多个机器人之间协作的日益复杂的任务。与传统方法相比，这些新条件需要更好地了解环境、机器人的位置以及施加给机器人的力和由机器人施加的力。嵌入式微处理器的功率越来越大，小型摄像机的质量越来越高，使用MEMS技术（微机电系统）的传感器（加速度计、陀螺仪等）的小型化，加上生产成本的降低，为这些技术在机器人应用中的应用和集成打开了大门，这些技术无法应对传统传感器的尺寸和重量。例如，移动机器人应该能够在平面上使用里程测量精确地沿着规定的路径导航。然而，机器人的精确几何形状、在车轮和地面之间滑动以及机器人在与障碍物碰撞后的重新定向等方面的不准确性将导致位置估计的误差越来越大。利用来自其他传感器的数据（加速度、角速度、视觉检测到的地标等），我们的目标是提供一种尚未达到的状态估计精度，因为与传统传感器相比，新型传感器的质量、尺寸和功耗都大幅降低。该项目的目标是开发机器人控制系统，充分利用最近在制造小型低成本传感器方面的技术突破。

Mobile robots must carry out increasingly complex tasks involving collaboration between two or more robots. These new conditions require a better knowledge of the environment, of the position of the robot,and of forces applied to and by the robot in a more precise manner than permitted by traditional approaches. The increasing power of embedded microprocessors, the higher quality of small video cameras, the miniaturization of sensors (accelerometers, gyro-meters, ...) using MEMS technologies (Micro-Electro-Mechanical Systems), combined with lower costs of production open the door to the utilization and integration of these technologies in robotics applications which could not cope with the size and weight of traditional sensors.For example, a mobile robot should be able to navigate on a planar surface with accuracy using odometric measurements to follow the prescribed path. However, inaccuracies about the exact robot geometry, sliding between the wheels and the ground, and reorientation of the robot after a collision with an obstacle will generate increasing errors in position estimates. Using data from additional sensors  (acceleration, angular velocity, landmarks detected by vision, ...), we aim to provide a level of accuracy not yet reached for state estimation given the huge reduction in mass, size and power consumption of new sensors compared to traditional sensors.The goal of this project is to develop robotic control systems taking full advantage of recent technical breakthroughs in the fabrication of small low cost sensors.