Soft robotic manipulators, unlike their rigid-linked counterparts, deform continuously along their lengths similar to elephant trunks and octopus arms. Their excellent dexterity enables them to navigate through unstructured and cluttered environments and to handle fragile objects using whole arm manipulation. This paper develops optimal designs for OctArm manipulators, i.e., multisection, trunklike soft arms. OctArm manipulator design involves the specification of air muscle actuators and the number, length, and configuration of sections that maximize dexterity and load capacity for a given maximum actuation pressure. A general method of optimal design for OctArm manipulators using nonlinear models of the actuators and arm mechanics is developed. The manipulator model is based on Cosserat rod theory, accounts for large curvatures, extensions, and shear strains, and is coupled to the nonlinear Mooney–Rivlin actuator model. Given a dexterity constraint for each section, a genetic algorithm-based optimizer maximizes the arm load capacity by varying the actuator and section dimensions. The method generates design rules that simplify the optimization process. These rules are then applied to the design of pneumatically and hydraulically actuated OctArm manipulators using 100psi and 1000psi maximum pressures, respectively.
Language-Guided Adaptive Perception for Efficient Grounded Communication with Robotic Manipulators in Cluttered Environments
