Optimization of Hybrid Power Sources for Mobile Robotics through the Use of Allometric Design Principles and Dynamic Programming

This work presents design rules and methods used to optimize mobile ground robots early in the design process. The focus ranges from the general geometric considerations for the overall size of a robot, to the optimization of size and control of three sources of hybrid power for specific ground robots. The geometric analysis includes study of the platform’s performance requirements for climbing, traversal and speed, and this work demonstrates that one can accurately calculate the necessary bulk properties of the robot including physical size, mass and power. Once the bulk properties are calculated, a system-level model can be designed for the robot platform using user-specified performance criteria. This system-level view decomposes the robot as a whole into its subsystems and the powertrain components used for locomotion. Such decomposition is used to accurately predict the necessary power, performance and allometry (size dependence) of each component. Once components have thus been correctly sized, the overall system-level performance is calculated including operational time and cruising distance. Comparisons to experiments on existing robot platforms show the fidelity of this approach. Comparisons between conceptual robot models at both system and subsystem-levels allow the user to examine tradeoffs between different performance requirements. In many cases, a specific sequence of tasks is needed for a robot to complete a given mission. This mission, for a given sized robot, translates into a power profile representing the power draw required to complete the task sequence. Dynamic Programming is used to optimize the control strategy and size of each component within the hybrid power source (battery, ultracapacitor and generator) for a number of missions. This work shows that, based on a given mission, the optimal power topology of a robot varies with the characteristics of its mission.