Abstract: The paper considers a novel modular and intrinsically safe redundant robotic system with biologically inspired actuators (pneumatic artificial muscles and rubber bellows actuators). Similarly to the biological systems, the stiffness of the internal parallel modules, representing 2 DOF joints in the serial robotic chains, is controlled by co-activation of opposing redundant actuator groups in the null-space of the module Jacobian, without influencing the actual robot position. The decoupled position/stiffness control allows the realization of variable joint stiffness according to different force-displacement relationships. The variable joint stiffness, as well as limited pneumatic muscle/bellows force ability, ensures internal system safety that is crucial for development of human-friendly robots intended for human-robot collaboration. The initial experiments with the system prototype demonstrate the capabilities of independently, simultaneously controlling both joint (Cartesian) motion and joint stiffness. The paper also presents the possible industrial applications of snake-like robots built using the new modules.
Abstract: A cognitive collaborative reinforcement learning
algorithm (CCRL) that incorporates an advisor into the learning
process is developed to improve supervised learning. An autonomous
learner is enabled with a self awareness cognitive skill to decide
when to solicit instructions from the advisor. The learner can also
assess the value of advice, and accept or reject it. The method is
evaluated for robotic motion planning using simulation. Tests are
conducted for advisors with skill levels from expert to novice. The
CCRL algorithm and a combined method integrating its logic with
Clouse-s Introspection Approach, outperformed a base-line fully
autonomous learner, and demonstrated robust performance when
dealing with various advisor skill levels, learning to accept advice
received from an expert, while rejecting that of less skilled
collaborators. Although the CCRL algorithm is based on RL, it fits
other machine learning methods, since advisor-s actions are only
added to the outer layer.