AI and Robotics at TRU: Institutional Augmentation – Part 2

CFJC was invited for a show-and-tell session on campus, with students offering project demonstrations for Episode 2 of AI and Robotics at TRU: Institutional Augmentation.

Gursahib Singh stood in front of a computer screen and wiggled his fingers, smiling proudly while a robotic hand mimicked his motions.

Singh, Yassh Singh and Deepansh Sharma used cameras, coding, voice AI, sensors and motors to Frankenstein an automated hand designed to improve efficiency in monotonous industrial work.

“Many people have worked on robotic arms,” Sharma said. “The main difference is they worked on one particular thing. One research paper worked on voice AI. One research paper worked on hand movements. One research paper tried to show it can be done. What we did differently was try to combine all of them and come up with one unique solution we can show industries.”

The project – part of the curriculum for Internet of Things, a semester-long Aighobahi offering – came together for less than $100.

“We apply artificial intelligence to objects,” Aighobahi said, offering a brief description of the course.

Gursahib Singh talked to the hand to demonstrate the voice AI element, asking it to divide 90 by 45. Index and middle fingers popped up.

“Anyone can come in front of a camera and use his hand to teach the robotic arm what to do and when to do it and then you can play it on a loop,” Gursahib Singh said. “All the motions from the hand, all the finger motions, all the joints that are there translate in a way that the machine understands which motors to move. Then it is co-ordinating and showing everything on the dashboard – which machine is working, how long it has been working, what is the energy consumption of that particular machine.”

Beside the automated hand on a table in front of Aighobahi was a digital-twin-plant system, another one of his student’s projects.

“If the plant is not getting enough water, you can see it visually from an application, press a button and it pumps water into the pot,” he said, noting the project took four months to complete. “If it’s not getting enough air, there’s a fan there, as well. It was very interesting and innovative. The pots, the fan, the tank, everything has to be represented in the app. That was the most challenging part and it was really fun to work with.”

The student exhibition continued with Fernando Ferrufino, whose project may be of interest to the BC Wildfire service.

“There’s a problem right now that wildfires are hard to detect and the way they’re detecting wildfires is via satellite mostly, so we wanted to find an alternative solution for this,” Ferrufino said.

Ferrufino held up solar-powered modules and sensors that collect data – reading temperature, humidity and gas – which is then crunched by AI for wildfire-detection purposes.

“The thing that makes it new and innovative is that we’re using backend networking that is very fast compared to wifi,” Ferrufino said, noting the modules and sensors would be strategically placed in forests. “It’s actually really, really fast. When we did the testing, it was giving 200 per cent more performance than satellite, for example.”

Added Aighobahi: “It can pick up the smoke and send a message to the control office and say, ‘Hey, there is a likelihood of fire in this area and it can predict it, as well.’ That’s where the artificial intelligence comes in.”

Graduate student John Bardelosa dodged volleyballs at the Tournament Capital Centre – home of the TRU WolfPack – while spilling on his masters project.

He is using video of WolfPack women’s volleyball matches to extract data on player locations and actions, and feeding statistical models to glean advantageous insight for coaches.

“Right now, we’re able to give live feedback from video recording, which is a good sign,” Bardelosa said. “It means our model is working and able to provide real-time feedback.”

Sean Hellingman, TRU assistant professor in math and statistics, is among the project supervisors.

“Thinking about where the tech is going in this area – artificial intelligence and computer vision – why don’t we just build something that’s going to collect this information for us?” Hellingman said.

Bardelosa is attempting to better the more expensive, less expeditious technology that’s out there.

“He’s extracting frames from the video and then he’s drawing boxes on what we’re interested in,” Hellingman said. “We’re teaching the model … you’re going to have a snapshot of if something went wrong or something went well, where everybody is, where the ball was and what the action was in that case.”

The hope is for cash-strapped collegiate and semi-pro teams to access an affordable platform through a user-friendly dashboard on a tablet.

“AI is a tool,” said Andrea Li, TRU special advisor to the president on artificial intelligence. “We’re able to process an exceptional amount of data in a fraction of a second. Our vision is start with the TRU community – students, researchers, educators and employees – so that we’re able to diffuse the ability to innovate across our community.”

Should Bardelosa pursue the project, he may introduce a conversational interface.

“You could ask the chatbot, ‘Hey, what’s the service efficiency of this player? Who do we target? Who has the lowest passing accuracy?’” Bardelosa said.

So you can expose the poor players on the opposition?

“Definitely,” Bardelosa said with a laugh. “That’s how you play the game.”

Next week, in Episode 3 of AI and Robotics at TRU, administration and faculty will discuss a plethora of ethics topics and their vision for the future of technology at the Kamloops university.

by Marty Hastings

AI and Robotics at TRU: Institutional Augmentation – Part 2

Our Community