Search for: All records

Cable driven parallel robots (CDPRs) are often challenging to model and to dynamically control due to the inherent flexibility and elasticity of the cables. The additional inclusion of online geometric reconfigurability to a CDPR results in a complex underdetermined system with highly non-linear dynamics. The necessary (numerical) redundancy resolution requires multiple layers of optimization rendering its application computationally prohibitive for real-time control. Here, deep reinforcement learning approaches can offer a model-free framework to overcome these challenges and can provide a real-time capable dynamic control. This study discusses three settings for a model-free DRL implementation in dynamic trajectory tracking: (i) for a standard non-redundant CDPR with a fixed workspace; (ii) in an end-to-end setting with redundancy resolution on a reconfigurable CDPR; and (iii) in a decoupled approach resolving kinematic and actuation redundancies individually. 

Abstract Mobile manipulators that combine base mobility with the dexterity of an articulated manipulator have gained popularity in numerous applications ranging from manufacturing and infrastructure inspection to domestic service. Deployments span a range of interaction tasks with the operational environment comprising minimal interaction tasks such as inspection and complex interaction tasks such as logistics resupply and assembly. This flexibility, offered by the redundancy, needs to be carefully orchestrated to realize enhanced performance. Thus, advanced decision-support methodologies and frameworks are crucial for successful mobile manipulation in (semi-) autonomous and teleoperation contexts. Given the enormous scope of the literature, we restrict our attention to decision-support frameworks specifically in the context of wheeled mobile manipulation. Hence, here, we present a classification of wheeled mobile manipulation literature while accounting for its diversity. The intertwining of the deployment tasks, application arenas, and decision-making methodologies are discussed with an eye for future avenues for research. 

Safe operations of autonomous mobile robots in close proximity to humans, creates a need for enhanced trajectory tracking (with low tracking errors). Linear optimal control techniques such as Linear Quadratic Regulator (LQR) and Model Predictive Control (MPC) have been used successfully for low-speed applications while leveraging their model-based methodology with manageable computational demands. However, model and parameter uncertainties or other unmodeled nonlinearities may cause poor control actions and constraint violations. Nonlinear MPC has emerged as an alternate optimal-control approach but needs to overcome real-time deployment challenges (including fast sampling time, design complexity, and limited computational resources). In recent years, the optimal control-based deployments have benefitted enormously from the ability of Deep Neural Networks (DNNs) to serve as universal function approximators. This has led to deployments in a plethora of previously inaccessible applications – but many aspects of generalizability, benchmarking, and systematic verification and validation coupled with benchmarking have emerged. This paper presents a novel approach to fusing Deep Reinforcement Learning-based (DRL) longitudinal control with a traditional PID lateral controller for autonomous navigation. Our approach follows (i) Generation of an adequate fidelity simulation scenario via a Real2Sim approach; (ii) training a DRL agent within this framework; (iii) Testing the performance and generalizability on alternate scenarios. We use an initial tuned set of the lateral PID controller gains for observing the vehicle response over a range of velocities. Then we use a DRL framework to generate policies for an optimal longitudinal controller that successfully complements the lateral PID to give the best tracking performance for the vehicle. 

A fully-constrained 𝑛−𝐷𝑂𝐹 cable-driven parallel robot (CDPR) has wrench closure if there are 𝑛+1 cables exerting positive tensions spanning the wrench space. However, the quality of wrench closure is often dependent on the geometric configuration of the supporting in-parallel chains of the CDPR. The reconfigurability endowed by adding in-chain kinematic and/or actuation redundancy to a conventional cable robot could greatly improve quality of the workspace. However, the status of various joints (active, passive or locked) affect the complexity of the systematic formulation and ultimate wrench-based analysis. Past efforts have tended to equilibrate the forces in these systems in such a way as to avoid kinematic redundancies. To this end, we formulate the kinematics of the redundant reconfigurable CDPR using matrix Lie group formulation (to allow ease of formulation and subsequent generalizability). Reciprocity (and selective reciprocity) permits the development of wrench analyses including the partitioning of actuation vs structural equilibration components. The total wrench set is greatly expanded both by the addition of kinematic redundancy and selective actuation/locking of the joints. The approach adopted facilitates the holistic determination of the true wrench polytope which accounts for the wrench contributions from all actuation sources. All these aspects are examined with variants of a 4-PRPR planar cable driven parallel manipulator (with varied active/passive/locked joints).