Using AI programming to make robots smarter

Using AI programming to make robots smarter

Dive into the world of computational problem-solving with RMIT's “Programming Autonomous Robots” course. How do RMIT students help make more intelligent robots? Read on to find out!

Have you ever wondered what goes into making robots intelligent? This is the exact type of computational problem solving that students tackle in RMIT’s Programming Autonomous Robots course, a part of the Master of Artificial Intelligence program. 

This course utilises a number of different robot models, including a NAO (programmable humanoid) robot and some AWS DeepRacer vehicles. Dr. Ginel Dorleon notes that the class carries out numerous activities, and “what we do depends on the robot. NAO can talk, carry things and move around, whereas the AWS vehicles have to be much more specific in their tasks. However, key to everything is Kinematics”

Kinematics describes the motion of points, bodies (objects), and systems of bodies (groups of objects) without considering the forces that cause them to move. In essence, Kinematics is concerned with only the instantaneous values of the robot's coordinates. “Students learn kinematics and parameters - how to make the robot move, then they program more range of motion. Some of the key components include degree of freedom, the number of joints on the robot and locomotion,” states Dr. Dorleon. 

Programming coordinates is a foundational step in path planning for intelligent robots, which involves various techniques including, but not limited to, reinforcement learning. This type of learning is a part of creating an autonomous agent and is defined as a system that can make decisions and act in response to its environment, independent of direct instruction by a human user.  

a hand typing on laptop with a robot reaching out from the screen

To help a robot to move around in a constantly changing environment (such as self-driving cars), in-class activities include placing obstacles on the floor and teaching a robot to avoid them. Students then create algorithms to further teach the robot to make those decisions by itself. Dr. Dorleon notes that “Reinforcement learning is such an important part of AI – what happens when you need to train a model without enough data sets? You use reinforcement learning.”

In the final class project student must form teams of 3 or 4 people and then choose from a number of programming topics. These include having the NAO kick a football to score a goal and using the DeepRacer vehicles to recognise a human crossing street, know what a stop sign is, and run on a specific track. 

Interestingly, AWS has an international competition called the DeepRacer Race, and of time of writing, a Master of Artificial Intelligence student team is ranked #1 in Vietnam.   

As Dr. Dorleon states, “We have students with the knowledge and capabilities to help today’s industry robots work in a smarter way. But autonomous vehicles will come to Vietnam at some point and our students are well positioned to take lead in this area when it happens.” 

The concepts, theories and practical experience that comes from programming autonomous robots all become relevant across multiple AI applications.  Do you want to learn fundamental AI principles, and have the opportunity to have some fun with robots while doing so? Check out what RMIT’s Master of Artificial Intelligence program can do for you!


Related news