Associate Professor at Binghamton University Shiqi Zhang has spent the last four years upgrading robotic dog models, first “training” one to respond better to leash guidance. Now, along with his team, Zhang has utilized AI tools to train a Unitree Go2 model robotic dog on a large language model (LLM) to follow vocal commands, as well as adapt and respond to the surrounding environment.
Zhang and his team asked seven legally blind study participants to use the Go2 to help guide them around a large indoor environment spanning multiple rooms. The Go2 fed participants descriptions of each room as it guided, helping each person better navigate through the area. Watch the video above for footage from the study.
“For this work, we’re demonstrating an aspect of the robotic guide dog that is more advanced than biological guide dogs,” said Zhang. “Real dogs can understand around 20 commands at best. But for robotic guide dogs, you can just put (AI tool) GPT-4 with voice commands. Then it has very strong language capabilities.”
