Viterbi researchers make disease-fighting robot


Photo of the four-legged robot in grass
The four-legged robot moves semi-autonomously and is designed to sanitize surfaces and help fight the coronavirus. (Photo Courtesy of Quan Nguyen)

A research team from the USC Viterbi School of Engineering has created a four-legged, semi-autonomous robot to help fight the spread of coronavirus. The robot has been named Legged Agile Smart Efficient Robot for Disinfection (LASER-D) and is able to move on a variety of terrain to sanitize surfaces with a chemical spray-based disinfectant. The machine was created as a research opportunity for master’s students, combining two previous systems: a disinfection robot and a quadruped robot. 

LASER-D can currently be controlled from a short distance away, with its vision system being overseen by an operator. The vision system serves as a way for operators to control movement while the robot is in development. The team is currently using a machine learning system to help the robot identify disinfected surfaces and said the robot’s visual perception is better than that of a human. LASER-D moves with “waypoints,” which are areas and points that are mapped and designated by the operators to guide the robot through an area.

Anthony Nguyen, a graduate student studying mechanical engineering who worked on the LASER-D project, said the pandemic was a big shock to everyone but that he was glad to have been part of a team that worked to provide solutions to the world’s new problem.

“As engineers, I think we really feel the need to tackle the current issues that are going on in the world and come up with innovative solutions,” Nguyen said. “I think the USC engineering department did a great job of that. And there’s so many opportunities here at USC and to be able to work on the first quadrant robot for disinfection was a pleasure and an honor for me.” 

The team, made up of seven graduate students, is headed by Dr. Satyandra K. Gupta and Dr. Quan Nguyen, both professors at Viterbi. Initially, Gupta’s lab was in the process of developing a UV disinfection robot, while Quan Nguyen’s lab had focused on quadruped robots. The new team includes Zhiwei Deng of the Ming Hsieh Department of Electrical and Computer Engineering, as well as Abhinav Pandey, Anthony Nguyen, Pornrawee Thonapalin, Ruiqi Wang, Tailun Liu and Yiyu Chen of the Department of Aerospace and Mechanical Engineering.

At the start of the pandemic, Gupta and Quan Nguyen’s labs had the idea to combine their projects to create a new, more mobile robot capable of non-hazardous disinfection. The team modified the disinfection system and attached it to a robot better equipped to move through classrooms to aid in disease prevention.

“They decided to combine the disinfection system from the wheeled robot onto a quadruped robot instead… [LASER-D] can traverse through classrooms, stairs and slippery slopes,” Anthony Nguyen said. 

The new team, specifically formed for this project, was composed of four sub-teams, each designated to work on a different part of the project: motion planning, simulation, vision and experimental. 

The robot’s movement is controlled from a short distance away and can disinfect mechanically. The benefits of making a four-legged robot rather than a wheeled one with a new disinfection system is that, by adding legs, the robot can access a kind of three-dimensional space by moving its different joints. Anthony Nguyen described this as the difference between a UV light scanning a two-dimensional surface versus a three-dimensional spray system that can move around an object. 

The vision team was in charge of implementing LASER-D’s vision system, which helped remote operators control the robot’s movement and disinfection tasks. 

Nguyen said the role of the simulation team was to create artificial environments, or world environments, to define the waypoints for the robot. The team is able to plan out the motion of the robot so when it’s time to send it in for disinfection, LASER-D knows how to move. If the robot is able to disinfect most areas, but certain areas are too tight or too close, the simulation team is able to define the area’s waypoints beforehand.

Pornrawee Thonapalin, a graduate student studying aerospace and mechanical engineering, said she worked on the motion planning team, creating the waypoints to guide the robot’s movement after mapping out the location. During research, Thonapalin was able to begin the mapping process, which is followed by the data being sent to the simulation team to program LASER-D’s movement. 

In tandem with the experimental team, the group in charge of building and physically experimenting with the robot, the sub-teams worked to combine each of the different systems for the LASER-D project to be successful. “I have some experience [going] to the lab. Sometimes we have to go to the lab, one day for one person, or as many as two people to working on it, and it’s very difficult to work remotely because we have to experiment,” Thonapalin said.

Nguyen said that the biggest challenge in working on the project was not being able to meet in person to work on the robot and having to navigate creating something physical in a virtual space. His role on the simulation team did not require him to experiment with the robot in person. A lot of the project involved screen sharing on Zoom and working through different time zones. 

“I would say the pandemic was ultimately the biggest challenge for us, but at the same time if it wasn’t for the pandemic we wouldn’t have this project,” Nguyen said. “So I’m thankful that I got to work on something that’s benefitting the world or just helping us figure out a solution to something that’s affecting us as a population.”

The robot is still being trialed, and while it has been tested on campus as a method of disinfection, it is not yet meant to be used on a larger-scale. In addition, the robot isn’t fully autonomous yet; Dr. Gupta said the robot is “teleoperated” and must be reasonably close to its operator to be controlled. It’s also limited to its single battery charge and cannot travel very far. 

Despite these limitations, the robot is still semi-autonomous, meaning LASER-D only has to be set to work and supervised by the operator remotely. The team is hopeful about the future autonomy of the robot. Nguyen described some possibilities in which the robot could be used to reduce pollution and pick up trash autonomously in a time when LASER-D’s successors could be used for projects outside of disinfection research.

“Robotics in general and AI, it’s such a fast-growing industry, and with this project, you can go in so many different directions,” Nguyen said. “I definitely see this project being used in the real world … it was a lot of trial and error, and at the same time I think that a lot of the people on our team were really motivated to do their jobs and get things done, so I guess that’s why we were so successful in making it work.”