In the smart connected bikes project, we aim to build a smart-cycling ecosystem that improves cycling safety and experience. In the densely populated Netherlands, cycling is the most efficient mode of mobility for short distances. With the growing usage of bikes, infrastructure maintenance is important for safer and an enjoyable cycling experience. Poor bike lanes have an adverse effect on not only the bike’s condition but also on the cycling comfort and the cyclist’s health. There is no existing method to pre-emptively evaluate the condition of the bike-lane.
The student is expected to use off-the-shelf sensors like stereo cameras, Inertial Measurement Unit (IMU) or accelerometers to measure the road condition, or classify “good” and “bad” roads. The hardware is chosen and assembled on the bike. The Region Of Interest (ROI) in the camera image is identified using conventional computer vision techniques. The ROI is then labelled using, say, the IMU data as either “good” / “bad”, or with a score to quantify the road quality. The camera should face forward (with possibly a slight tilt downwards) to cover the road. The student is open to choose any method they prefer, but it should be computationally efficient. Knowledge of working with sensors and C++/Python is desired.
20% Theory, 60% Practical, 20% Writing
Deepak Yeleshetty – email@example.com