A new approach to designing healthcare robots flips the traditional model on its head: instead of engineers building robots and hoping clinicians will use them, hospitals are now inviting nurses, doctors, patients, and even artists to help design the robots themselves. A Cornell Tech-led study presented at the 2026 Association for Computing Machinery CHI Conference on Human Factors in Computing Systems documented a 14-week collaborative project where 22 participants from diverse backgrounds worked together to imagine and physically build robots that could genuinely ease daily burdens in healthcare settings.

What Happens When You Let Healthcare Workers Design Their Own Robots?

The traditional approach to healthcare robotics often starts with what engineers think is possible, then asks clinicians to adapt their workflows around the technology. This Cornell project reversed that logic entirely. Rather than beginning with technical capabilities, the team examined what actually frustrates healthcare workers, what confuses or stresses patients, and where a robot might realistically help. The result was a framework for what researchers call "considerate embodied AI".

Over three months, the 22 participants met weekly in Cornell Tech's MakerLAB, a hands-on learning environment dedicated to creative exploration and collaborative making. The teams tackled challenges across three real healthcare environments: an emergency department, a sleep disorder clinic, and a long-term rehabilitation facility. They moved step by step from brainstorming to cardboard mockups to full-size, interactive robot prototypes.

"Many healthcare facilities experience challenges managing and caring for patients, yet limited research explores the common issues faced by healthcare workers and patients, and how robot design could help," said Angelique Taylor, the Andrew H. and Ann R. Tisch Assistant Professor at Cornell Tech and a faculty fellow in the Cornell Institute for Healthy Futures.

Angelique Taylor, Assistant Professor at Cornell Tech

The physical prototyping process revealed insights that interviews and sketches alone never would have surfaced. When teams began constructing full-scale robots, practical issues like hallway width, patient comfort, noise levels, hygiene requirements, and safety suddenly became impossible to ignore. One emergency department team designed a bear-shaped robot that could deliver medical kits directly to patient rooms before a doctor arrives, saving precious time for nurses in high-pressure situations. A sleep clinic team built a gentle, concierge-style robot with calming lights to guide patients through unfamiliar nighttime procedures. For long-term care rehabilitation, participants focused on social connection, creating a robot that could provide entertainment, display daily schedules, and help residents feel less isolated.

How Does Co-Design Actually Change What Robots Can Do?

The MakerLAB served as what researchers called a "third place," separate from clinical hierarchies and academic pressure. This neutrality mattered enormously. As participants learned how robots move, sense, and interact, they grew more confident contributing ideas, even without technical backgrounds. Artists shaped how robots looked and felt. Long-term care residents flagged moments where machines might feel intrusive or dehumanizing. Healthcare workers identified workflows that technology often ignores entirely.

"Digital fabrication is not just a manufacturing step. It acts as an instrument of thought. It allows non-technical stakeholders to move from being passive observers to making grounded, expert judgments about complex AI," explained Niti Parikh, director of learning spaces and MakerLABs at Cornell Tech and founder of CRAFT@Large.

Niti Parikh, Director of Learning Spaces at Cornell Tech

Across all three healthcare settings, researchers found that robots were most valuable when they handled repetitive, non-clinical tasks. This freed human workers to focus on care that requires empathy, judgment, and personal connection. The robots weren't designed to replace nurses or doctors; they were designed to eliminate the tedious work that keeps clinicians from doing what they do best.

Steps to Implement Co-Design in Healthcare Robotics

• Create a Neutral Space: Establish a dedicated environment like a MakerLAB where clinical hierarchies dissolve and experimentation feels safe, allowing nurses, doctors, patients, and engineers to collaborate as equals.
• Start With Problems, Not Solutions: Begin by examining what frustrates healthcare workers and stresses patients, rather than asking what robots can technically do, ensuring designs address real clinical needs.
• Build Physical Prototypes Early: Move quickly from sketches to full-scale, interactive mockups so participants can identify practical constraints like hallway width, noise levels, and hygiene requirements that emerge only in physical space.
• Include Diverse Perspectives: Involve artists, long-term care residents, nurses, doctors, engineers, and computer scientists so that robots are designed for human experience, not just technical efficiency.
• Focus on Repetitive Tasks: Design robots to handle non-clinical, repetitive work like delivery, packaging, or scheduling, freeing clinicians to focus on care requiring empathy and personal judgment.

The findings come at a critical moment. Hospitals nationwide are looking to technology to address staffing shortages, burnout, and rising patient loads. But many healthcare facilities have deployed robots that actually complicate workflows rather than simplify them. This research suggests a better path forward.

"The physical prototype becomes the common language that connects a nurse's lived experience with a researcher's technical goals. Our work transformed participants into active co-designers, allowing them to physically externalize their expertise into functional, interactive prototypes," noted Parikh.

Niti Parikh, Director of Learning Spaces at Cornell Tech

The research team included co-authors Yuanchen Bai and Ruixiang Han, doctoral students in information science at Cornell Tech, and Wendy Ju, professor at Cornell Tech and the Cornell Ann S. Bowers College of Computing and Information Science. Their work suggests that as embodied AI systems become more prevalent in healthcare, the design process itself matters as much as the technology. Robots that are "considerate" of environmental context, patient conditions, and human needs are far more likely to be adopted and to actually improve care.

This approach reflects a broader shift in how physical AI is being deployed across industries. Rather than treating robots as standalone tools, forward-thinking organizations are recognizing that embodied AI systems must be designed with the people who will use them, not for them.