Coordinated Control Based On Vision Sensor For Mobile Robot In Rough Terrain

With the development of modern technology, research of robots is widely concerned.Especially, mobile robots and robot vision became a hotspot of research. This subject issupported by Project of National Natural Science Foundation of China, named “Trafficabilitycontrol with low-power consumption toward wheeled robot in rough terrain”. This thesismainly focuses on the building of visual feedback system and cross-coupling coordinatedcontrol system of multi driving wheels.At first, according to the research status and the tendency of visual servo and robotcontrol system, design a real-time vision measuring system and a coordinated control system.As the upper computer, the industrial PC motherboard GENE-QM57is chosen, which canrapidly measure the velocity of the mobile robot. Then, the IPC with the PC/104bus isselected as the lower computer, with the CAN bus for driving of motor and the steering servocontrol card for steering controlling. On that basis, the cross-coupling coordinated controlsystem of multi driving wheels can be done with the help of IMU, encoder and other sensors.Secondly, a monocular/binocular vision system is built for real-time measuring thevelocity of the mobile robot in various kinds of terrain, and the extracting and matching ofregion feature is provided, which works better than the matching methods based on featurepoint. The transformation of the coordinates in the vision system has been achieved, toestimate the velocity of the robot. According to different motion environment, a simplemonocular vision system is adopted in the flat ground, the binocular calibration system willbe used while the robot moves in the rough terrain to correct distance parameterz cof thevision system.Then, according to the existing robot platform, the coordinate transformation method isapplied to establish and analyze kinematics model, and this paper presents an algorithm forvisual estimation of wheel-ground contact angle on uneven rigid terrain. In addition, with thehelp of a fifth wheel is equipped on the platform, based on the established kinematics modelof the fifth wheel system, the slippage estimation method is derived combining the kinematicsmodeling of the robot. Combined with measured velocity information, a cross-couplingcoordinated control system model is proposed and chooses appropriate cross-coupling gainparameters, to reduce orientation and path error. Finally, the experiment of robot motion on hard ground and sandlot is done, includingthe vision measure system and the cross-coupling coordinated control system, the velocity ofeach wheel, the velocity of robot and the slippage information are fed back and analyzing.The experiment also tests the stability and effectiveness of the vision system and motioncontrol system.