CHANNELS

Aerospace

Automotive

Automated & Autonomous Vehicles

Battery & Electrification Technology

Defense

Drone TV

Electronics

Manufacturing & Prototyping

Materials

Medical

Motion Control

RF & Microwave Technology

Robotics & Automation

Sensors & IoT

Test & Measurement

Aerospace
Automotive
Automated & Autonomous Vehicles
Battery & Electrification Technology
Defense
Drone TV
Electronics
Manufacturing & Prototyping
Materials
Medical
Motion Control
RF & Microwave Technology
Robotics & Automation
Sensors & IoT
Test & Measurement
Aerospace & Defense
Search Results

Human-Robot Tandem Race: Programming Robots to Follow You

Master’s students from EPFL  participated in a race where the competitors were tandems of students and robots. The challenge was part of EPFL's "Data and AI for Transportation" class, and put the students’ programming skills to the test in pursuit of enhanced methods of human-machine interaction. The students positioned themselves at the starting line, holding up signs with visual signals designed to get their robots to follow them. The algorithms that the students developed are similar to those used in self-driving cars, which enable the vehicles to recognize street signs, traffic lights, pedestrians, and other cars.

Topics:
Automation Automotive Motion Control Robotics Software

More From SAE Media Group

    Transcript

    00:00:00 Our main motivation is to develop robots that can coexist with humans. And the main challenge is how they can coexist without affecting human behavior, but also respect social conventions and ethical rules. And this race is just an opportunity to show that we can have humans and robots coexist even in a race, but then on the research side, we want to show that we can have the same technology in crowded environment. So, the idea is that the students are going to build their own perception module for the robots, and then the robot can use those algorithms to track a particular object that they want the robot to track,

    00:00:42 and then, they lead the robot moving in the environment. It was fun in the beginning to see how the robots didn't track our image at all, and then, when we started training it, when we started improving our model, the robot then tracked it better and better, and had less and less false detections. We trained and network to recognize us however, this approach is very error-prone, since during the competition, there are a lot of people competing at the same time, so the robots will be distracted by all the other people. A better idea is to keep an object that is easily recognizable,

    00:01:20 and train the network on this object. We trained our model to recognize a specific image, a pattern and a color to track it and follow it wherever we take it within the camera angle. So the robot recognizes the image of Mickey Mouse, follows it and tracks it. Today the race was amazing, we had a very good time making the code for our robot, and our detection system was working better than we thought, it was good. We did a little bit of coding we have some background but very very little, because we're from civil engineering so we were not advantaged compared to some other colleagues from informatics section and others, so it was really the first time we did a code from A to Z. Neural networks, artificial intelligence, and data in the future

    00:02:09 will become more and more important, so for every engineer, having a basis of this stuff will certainly be very useful when using these things in practice in the future.