Adversarial Generation of Informative Trajectories for Dynamics System Identification
This addresses a bottleneck in robotics for system identification by making trajectory generation more efficient and scalable.
The paper tackles the problem of expensive and intractable trajectory generation for dynamic system identification in robotics by proposing multiple shorter cyclic trajectories generated in parallel and combined, scaled further with GAN-based architectures. The result is dramatic acceleration of trajectory optimization and more accurate system identification compared to conventional approaches.
Dynamic System Identification approaches usually heavily rely on the evolutionary and gradient-based optimisation techniques to produce optimal excitation trajectories for determining the physical parameters of robot platforms. Current optimisation techniques tend to generate single trajectories. This is expensive, and intractable for longer trajectories, thus limiting their efficacy for system identification. We propose to tackle this issue by using multiple shorter cyclic trajectories, which can be generated in parallel, and subsequently combined together to achieve the same effect as a longer trajectory. Crucially, we show how to scale this approach even further by increasing the generation speed and quality of the dataset through the use of generative adversarial network (GAN) based architectures to produce a large databases of valid and diverse excitation trajectories. To the best of our knowledge, this is the first robotics work to explore system identification with multiple cyclic trajectories and to develop GAN-based techniques for scaleably producing excitation trajectories that are diverse in both control parameter and inertial parameter spaces. We show that our approach dramatically accelerates trajectory optimisation, while simultaneously providing more accurate system identification than the conventional approach.