Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces CrashPBO, an extension of Preferential Bayesian Optimization (PBO) that allows users to provide both pairwise preference feedback and explicit crash reports during black-box optimization. By incorporating crash feedback, CrashPBO avoids exploring parameter regions that frequently lead to failures, improving data efficiency and reducing hardware damage. Experiments on synthetic functions and three robotic platforms (quadcopter, quadruped, and manipulator) demonstrate that CrashPBO reduces crashes by 63% and enhances the practicality of PBO in real-world robotics applications.
CrashPBO lets you teach robots new tricks faster and with fewer crashes, by directly incorporating "this setting broke the robot" feedback into Bayesian optimization.
Bayesian optimization is a popular black-box optimization method for parameter learning in control and robotics. It typically requires an objective function that reflects the user's optimization goal. However, in practical applications, this objective function is often inaccessible due to complex or unmeasurable performance metrics. Preferential Bayesian optimization (PBO) overcomes this limitation by leveraging human feedback through pairwise comparisons, eliminating the need for explicit performance quantification. When applying PBO to hardware systems, such as in quadcopter control, crashes can cause time-consuming experimental resets, wear and tear, or otherwise undesired outcomes. Standard PBO methods cannot incorporate feedback from such crashed experiments, resulting in the exploration of parameters that frequently lead to experimental crashes. We thus introduce CrashPBO, a user-friendly mechanism that enables users to both express preferences and report crashes during the optimization process. Benchmarking on synthetic functions shows that this mechanism reduces crashes by 63% and increases data efficiency. Through experiments on three robotics platforms, we demonstrate the wide applicability and transferability of CrashPBO, highlighting that it provides a flexible, user-friendly framework for parameter learning with human feedback on preferences and crashes.