| Abstrakt: |
Selecting suitable kinematic configurations for autonomous, flexible mobile systems remains a prominent challenge in various applications of trajectory tracking, including path solver planning, redundant robot control, grasping, and robotic system design. Current techniques based on the inverse kinematic problem suggest that the classic manipulability index referencing the local frame’s velocity is sufficient to infer the behavior of the kinematic constraints. Successful constraint inference then allows the control strategy to be updated when the performance becomes suboptimal with respect to the target path. However, these approaches become deficient when the joint space is flexible, making it necessary to explicitly account for the structural parameters of the mobile system (e.g., number of joints, mass, wheel size, and links) in order to infer the behavior of the kinematic constraints. Instead, the dynamic confined space of velocities (DCSV), initially proposed by the author, suggests a Markovian process relating the local frame’s velocity, kinematic constraint violations, and implicit flexibility. Hence, the present paper analyzes the DCSV’s behavior and proposes an offline method to reconfigure the structural parameters, thus demonstrating how the DCSV’s behavior can be used to reduce the violation of kinematic constraints. To this end, two lemmas and one theorem provide a new viewpoint regarding manipulability by using invariant and ergodic measurement conditions corresponding to the kinematic constraint evolution to compose DCSV-based selection criteria. The results are based on tracking problems and path-planning solvers in which the manipulability of a kinematic configuration is improved by updating the configuration according to optimal paths. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink