Abstract
We address the problem of enabling quadrupedal robots to perform preciseshooting skills in the real world using reinforcement learning. Developingalgorithms to enable a legged robot to shoot a soccer ball to a given target isa challenging problem that combines robot motion control and planning into onetask. To solve this problem, we need to consider the dynamics limitation andmotion stability during the control of a dynamic legged robot. Moreover, weneed to consider motion planning to shoot the hard-to-model deformable ballrolling on the ground with uncertain friction to a desired location. In thispaper, we propose a hierarchical framework that leverages deep reinforcementlearning to train (a) a robust motion control policy that can track arbitrarymotions and (b) a planning policy to decide the desired kicking motion to shoota soccer ball to a target. We deploy the proposed framework on an A1quadrupedal robot and enable it to accurately shoot the ball to random targetsin the real world.