Desk navigating robot using computer vision
Computer Science (Games) BSc(Hons) student John Evans tells us a bit about his course and demos his final year project robot at the Computing Degree Show.
“I chose this course as it is a general computer science course which could give me more options compared to focusing purely on games. The artificial intelligence modules and web-based games module were the most exciting for me personally, as I got to explore interesting concepts with technologies I enjoyed using.
The teaching staff were very knowledgeable and were able to give great one to one advice about particular topics in the modules they taught, or even about uni in general.
The most challenging modules of my course were the more maths heavy modules like the Games Development C++ modules, but in the wider sense of being at university, time management was probably my biggest challenge! The most interesting parts were learning both the very high-level programming techniques such as neural networks and lambda functions, and the very low level (syntax etc.) knowledge I gained that has allowed me to learn several different languages.
My ‘Desk Navigating Robot Using Computer Vision’ final year project explores the feasibility of using cost-effective camera-based computer vision to create a robot that can move around a desk, avoiding falling off the edge or running into walls.
Similar problems are often faced within the fields of computer science and robotics and are usually solved via either a) expensive Lidar and/or environmental mapping systems, or b) Infrared or ultrasonic sensors which are comparatively cheap but can give unreliable or slow results. Thanks to recent hardware and software advancements, using computer vision to replace or supplement those more established methods has become more viable.
The general problem-solving, programming, and research experience I gained on my course will help me whatever I decide to do in the future.