Spring Robotics Colloquium: Tapomayukh Bhattacharjee (Cornell)

Physical robot caregiving requires balancing task performance with the specific, multifaceted needs of users, including those with severe mobility limitations. Effective systems must integrate human factors—such as cognitive workload and engagement levels—directly into algorithmic decision-making rather than treating humans merely as passive recipients. Because caregiving environments are highly variable, robots benefit from modular hardware and software that allow for on-the-fly personalization and transparent interaction. Key strategies include using foundation models for high-level reasoning while maintaining layered safety guardrails that range from software-based constraints to hardware-level fail-safes. Ultimately, the goal is to augment human caregiving by providing reliable, adaptable support that respects user autonomy and dignity, moving beyond rigid, one-size-fits-all automation toward systems that learn from subtle, real-world interactions and prioritize the specific preferences of the individual receiving care.