轮腿式六足机器人步态规划及姿态融合算法研究

发布时间：2018-10-15 12:48

【摘要】：在移动机器人中,最常见的两类机器人是轮式机器人和腿式机器人。轮式机器人是应用范围最广,在平面上可靠性最高的机器人,缺点是对地面环境的适应力差,不能应用于不平整地形。相对于轮式机器人,腿式机器人的足点分布更加灵活,所以可以适应更复杂的环境,可以在沙地、草地等不平整地形稳定的行走,缺点是能量利用率低,速度缓慢。为了使机器人同时拥有腿式机器人对地面适应力强和轮式机器人在平面上运行速度快、能量消耗低的优点,本文设计了混合运动模式的轮腿式六足机器人。本文搭建了被动轮与腿融合的轮腿式六足机器人,分别对摆动腿和支撑腿进行了运动学分析。并利用多目标粒子群算法,以最大步距和最大步高为目标,优化了机器人的初始位置,使机器人的运动性能可以兼顾最大步高和最大步距两个因素。设计了基于正弦线的足端轨迹和用于行走的步态规划方法,并对前进三足步态和原地旋转三足步态做了具体分析。同时依据轮滑原理,设计了用于机器人滑行的步态,并对规划的参数进行了优化。在基于ROS的Gazebo仿真平台下,对所设计的机器人进行建模,并对设计的步态进行仿真验证。实验结果表明,机器人可以成功的执行前进步行和前进滑行步态,关节角度光滑无突变。为了对机器人进行方向控制,本文基于多传感器的信息融合解算了机器人的姿态角。利用启发式漂移消减法(HDR)对陀螺仪的输出信号滤波,得到了比较准确的机体角速度值。分别利用陀螺仪和磁力计配合加速度计解算机器人的姿态角,然后应用卡尔曼滤波算法,融合陀螺仪动态性能好和磁力计静态性能好的特点,解算出了准确的机器人姿态角。搭建物理实验平台,应用NDI光学定位系统测取机器人的位置信息。利用搭建的姿态解算模块,测量机器人的方向角。在搭建的实体样机上进行物理实验,可以成功执行前进步行、前进滑行、原地旋转步行、原地旋转滑行步态,达到了预期的效果。
[Abstract]:Among mobile robots, the two most common robots are wheeled robots and legged robots. Wheeled robot is the most widely used robot with the highest reliability on the plane. Its shortcoming is that it can not be applied to uneven terrain because of its poor adaptability to the ground environment. Compared with wheeled robots, legged robots are more flexible in the distribution of foot points, so they can adapt to more complex environment and walk stably in sandy land, grassland and other uneven terrain. The disadvantages are low energy utilization ratio and slow speed. In order to make the robot have the advantages of high adaptability to the ground and high speed and low energy consumption of the wheeled robot on the plane, a wheel-legged six-legged robot with mixed motion mode is designed in this paper. In this paper, a wheel-leg six-legged robot with the fusion of passive wheel and leg is built, and the kinematics analysis of swing leg and supporting leg is carried out respectively. Based on the multi-objective particle swarm optimization algorithm, the initial position of the robot is optimized by taking the maximum step distance and the maximum step height as the target, so that both the maximum step height and the maximum step distance can be taken into account in the motion performance of the robot. The foot trajectory based on sinusoidal line and the gait planning method for walking are designed, and the forward tripodal gait and the in-situ rotating tripodal gait are analyzed in detail. At the same time, according to the roller skating principle, the gait used for robot gliding is designed, and the planning parameters are optimized. In the Gazebo simulation platform based on ROS, the robot is modeled and the gait is simulated. The experimental results show that the robot can successfully perform forward walking and forward sliding gait, and the joint angle is smooth without mutation. In order to control the direction of the robot, the attitude angle of the robot is calculated based on multi-sensor information fusion. The output signal of gyroscope is filtered by heuristic drift subtractive method (HDR). The attitude angle of the robot is calculated by using gyroscopes and magnetometers with accelerometers respectively. Then Kalman filtering algorithm is applied to combine the characteristics of gyroscopes with good dynamic performance and static performance of magnetometers to solve the accurate attitude angle of the robot. The physical experiment platform is built, and the position information of robot is measured by NDI optical positioning system. The orientation angle of the robot is measured by using the constructed attitude solution module. The physical experiments are carried out on the physical prototype, which can successfully execute forward walking, forward gliding, in-situ rotation walking, and in-situ rotation taxiing gait, which achieves the desired effect.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】