A modular approach to the dynamics of complex robot systems

The need for a more general means of modelling and simulating the dynamics of arbitrary robot systems has developed from the increased complexity of modern robot tasks. Today, cooperating robots are used in unstructured environments such as hazardous waste remediation and space-based construction. Dynamic simulation of such systems is complicated by the numerous contacts between the many system structures. Most current simulation algorithms are not well suited to the treatment of multiple contacts and require substantial reprogramming when new system tasks or configurations occur.; This dissertation presents a Modular Robot Dynamic Simulation (MRDS) algorithm, which is capable of handling serial, parallel, and hybrid series/parallel topologies for both open-chain and closed-chain, multi-robot systems. Modularity allows the open chain dynamics of all system structures (individual robots, payloads, devices, etc.) to be determined independently and simultaneously. Consequently, parallel processing is ideal, as one module may be treated per processor resulting in increased computational efficiency and real-time capability.; The algorithm permits contacts between modules to be holonomic, non-holonomic, constant, or time-varying. Special attention is paid to the series connection of modules, as this ability to connect objects dynamically without the need to rederive and re-code new equations or coupling terms is new for complex robot dynamic simulation. The MRDS algorithm is also applicable to constrained, structurally flexible manipulators. Second order strain and kinematic effects are discussed, including appropriate linearization schemes for both open and closed loop topologies. Validation of the algorithm is achieved through the simulation of various multi-module, structurally flexible systems including parallel cooperations, serial cooperations, base-excitation, and macro/mini formulations.; The MRDS algorithm is especially well suited for the constrained motion dynamic simulation of complex, multi-robot, reconfigurable systems. The modular nature allows alternate configurations of the same system components to be simulated quickly and easily. The MRDS algorithm is also computationally efficient, open-ended, and general, making it particularly valuable for the treatment of the increasingly more complex robot systems in use today.