MARL Robots: Simulating Robots in Emergency Rooms

healthcare simulator

Research Question: How can we ensure seamless and safe interactions between robots and humans in healthcare workplaces by simulating their interactions using online programs?

I'm working with PhD candidate Promise Ekpo, mentored by Professor Angelique Taylor as part of Cornell Tech's AIRLab. Promise built an online simulator of an emergency room, and we have been helping her run various experiments using the simulator to figure out how to optimize the workflows and task assignments of healthcare workers, subject to various constraints. I'm collaborating with another undergraduate student, Saesha Agarwal, on a sub-project to introduce delivery robots to this simulator, with two main goals: to find an optimal way of incorporating robots into the existing workflows, and to determine whether emergency rooms are more or less efficient when robots are introduced to the setting.

Student: Arshia Agrawal

Collaborators: Promise Ekpo, Saesha Agarwal

From Promise Ekpo: Arshia's project was focused on hyperparameter optimisation of an existing RL simulator project led by Promise Ekpo and other collaborators in the Airlab at Cornell Tech. Through running experiments, she discovered the importance of hyperparameter tuning on new RL environments, as several RL algorithms performed better after hyperparameter tuning.

Top