On Robot Compliance: A Cerebellar Control Approach

The work presented here is a novel biological approach for the compliant control of robotic arm in real time (RT). We integrate spiking cerebellar network at core feedback loop performing torque-driven control. controller provides torque commands allowing accurate and coordinated movements. To compute these output motor commands, receives robot's sensorial signals, goal behavior, an instructive signal. These input signals are translated into set evolving patterns representing univocally specific system state every point time. Spike-timing-dependent plasticity (STDP) then supported, building adaptive continuously adapts provided to robot from experience as STDP deployed. Adaptive turn, help cope with built-in elastic elements within actuators mimicking human muscles (inherently elastic). propose natural integration bioinspired scheme, based on cerebellum, robot. prove that our outperforms accuracy default factory-installed position tasks used addressing behavior: controlling six degrees freedom (DoF) smooth movements, fast ballistic unstructured scenario