Deploying large, complex policies in the real world requires the ability to steer them to fit the needs of a situation. Most common steering approaches, like goal-conditioning, require training the robot policy with a distribution of test-time objectives in mind. To overcome this limitation, we present DynaGuide, a steering method for diffusion policies using guidance from an external dynamics model during the diffusion denoising process. DynaGuide separates the dynamics model from the base policy, which gives it multiple advantages, including the ability to steer towards multiple objectives, enhance underrepresented base policy behaviors, and maintain robustness on low-quality objectives. The separate guidance signal also allows DynaGuide to work with off-the-shelf pretrained diffusion policies. We demonstrate the performance and features of DynaGuide against other steering approaches in a series of simulated and real experiments, showing an average steering success of 70% on a set of articulated CALVIN tasks and outperforming goal-conditioning by 5.4x when steered with low-quality objectives. We also successfully steer an off-the-shelf real robot policy to express preference for particular objects and even create novel behavior. Videos and more can be found on the project website: https://dynaguide.github.io

For middle-school math students, interactive question-answering (QA) with tutors is an effective way to learn. The flexibility and emergent capabilities of generative large language models (LLMs) has led to a surge of interest in automating portions of the tutoring process - including interactive QA to support conceptual discussion of mathematical concepts. However, LLM responses to math questions can be incorrect or mismatched to the educational context - such as being misaligned with a school's curriculum. One potential solution is retrieval-augmented generation (RAG), which involves incorporating a vetted external knowledge source in the LLM prompt to increase response quality. In this paper, we designed prompts that retrieve and use content from a high-quality open-source math textbook to generate responses to real student questions. We evaluate the efficacy of this RAG system for middle-school algebra and geometry QA by administering a multi-condition survey, finding that humans prefer responses generated using RAG, but not when responses are too grounded in the textbook content. We argue that while RAG is able to improve response quality, designers of math QA systems must consider trade-offs between generating responses preferred by students and responses closely matched to specific educational resources.

Drilling a hole on a curved surface with a desired angle is prone to failure when done manually, due to the difficulties in drill alignment and also inherent instabilities of the task, potentially causing injury and fatigue to the workers. On the other hand, it can be impractical to fully automate such a task in real manufacturing environments because the parts arriving at an assembly line can have various complex shapes where drill point locations are not easily accessible, making automated path planning difficult. In this work, an adaptive admittance controller with 6 degrees of freedom is developed and deployed on a KUKA LBR iiwa 7 cobot such that the operator is able to manipulate a drill mounted on the robot with one hand comfortably and open holes on a curved surface with haptic guidance of the cobot and visual guidance provided through an AR interface. Real-time adaptation of the admittance damping provides more transparency when driving the robot in free space while ensuring stability during drilling. After the user brings the drill sufficiently close to the drill target and roughly aligns to the desired drilling angle, the haptic guidance module fine tunes the alignment first and then constrains the user movement to the drilling axis only, after which the operator simply pushes the drill into the workpiece with minimal effort. Two sets of experiments were conducted to investigate the potential benefits of the haptic guidance module quantitatively (Experiment I) and also the practical value of the proposed pHRI system for real manufacturing settings based on the subjective opinion of the participants (Experiment II).