A Solution to Adaptive Mobile Manipulator Throwing

Mobile manipulator throwing is a promising method to increase the flexibility and efficiency of dynamic manipulation in factories. Its major challenge is to efficiently plan a feasible throw under a wide set of task specifications. We analyze the throwing problem and show that it can be reduced to a simpler planar problem, hence reducing greatly the computational costs. Using data analysis and machine learning, we build a model of the object's inverted flying dynamics and the robot's kinematic feasibility, which enables throwing motion generation in 1 ms given target position query. Due to the computational efficiency of our method, we show that, the system is adaptive when disturbed during task execution, via replanning on the fly to find out an alternative throw, instead of sticking to the original plan. Code is available at: https://github.com/liuyangdh/mobile-throwing