Long-range throwing is a useful skill for mobile robots in delivery and retrieval, but it demands high release velocity, coordinated motion, and resilience to terrain disturbances. The authors propose a whole-body kinetic-coordination framework for a wheeled-quadruped manipulator, combining a kinetic-chain arm model with SQP-based trajectory optimization and a reference-guided reinforcement learning policy for base motion.
When the arm is optimized on a stationary base, it reaches a release speed of 12.75 m/s and a throwing distance of 16.68 m. Activating whole-body coordination raises these figures to 16.64 m/s and 28.25 m, showing that coordinated base movement substantially extends throwing performance.
The team then integrated the platform's mobility and validated the full policy in hardware experiments across flat ground, grass, slopes, and rough roads. The results indicate that kinetic-chain optimization plus whole-body coordination enables reliable long-range throwing over diverse unstructured terrains. This single-source study does not present competing results, so the reported gains stand as the paper's main claim.