A Generalized Kinematic Synthesis and Optimisation Framework for Serial Manipulator Robots in Cluttered Environments
Finding optimum kinematic designs for non-standard robotic manipulators, such as those used in medical, nuclear, and space applications is challenging due to the need to adapt to complex tasks within constrained environments. This design optimization problem is multi-dimensional and non-convex, with nonlinear constraints. Ensuring reachability, i.e., the existence of continuous trajectories between required positions for serial articulated manipulators in the presence of obstacles, adds another layer of complexity. Traditional approaches often rely on standard motion planning or inverse kinematics methods incorporated directly into the kinematic design optimization loop, resulting in considerable computational expense. To address these challenges, we introduce a new, robust design framework grounded in a generalized parametric kinematic model. This framework enables simultaneous optimization of both the robot’s link dimensions and its kinematic topology for arbitrary tasks within constrained environments, enhancing both computational efficiency and the quality of the resulting designs. The effectiveness of the proposed design framework is verified and evaluated through comparisons with baseline benchmarks. Results demonstrate that the novel design framework, by using Genetic Algorithm, can accelerate kinematic design optimization compared to the current state-of-the-art and optimize link dimensions and joint types simultaneously for serial robots operating in cluttered environments.