Robots avoid dead ends using lidar and AR

The ability of humans to avoid blind alleys is thanks to our sensing of appropriate signs and assessing distances. Our perception allows us to move back onto the correct path with little time wasted turning back – a skill that cannot be taken for granted in robots.

However, this is exactly the behaviour programmed into robots by researchers from Nicolaus Copernicus University’s Institute of Engineering and Technology.

Despite robots being largely used in predictable, organised environments such as warehouses and factories, they can still find their path blocked. According to Professor Tomasz Tarczewski, their main problem is running into dead ends, causing them to enter and exit unhelpful paths, wasting energy and time. One way to solve this problem is to implement local path-planning algorithms to ease avoidance of obstacles, making it possible for the robot to unblock itself autonomously and continue with its task.

This required Tarczewski and colleagues to introduce a lidar sensor to their robots. Lidar sensors, used widely in technologies such as autonomous vehicles, use reflected pulses of laser light to measure the distance between a subject and objects in its environment.

This information feeds into a “dead-end prediction mechanism”; the robot compares real-time lidar information about its surroundings with information about its own dimensions and makes a decision about whether it can pass in a certain direction without becoming trapped. In order for a robot to give up exploring a certain path, it must be provided by a mechanism to bypass it.

For this task, the researchers applied augmented reality (AR) to create a “virtual wall” – along with additional obstacles to the left and right – in the path of the robot, preventing it from passing into a dead end and instead guide it in the right direction.

This combination of path-planning algorithm, lidar and AR forms the basis of the Nicolaus Copernicus University researchers’ platform for preventing robots becoming stuck.

The team of researchers implemented the platform in a mobile robot (Husarion ROSbot 2.0 PRO) and conducted a series of laboratory tests to demonstrate the effectiveness of the proposed solution. According to a statement from the university, the results are “very promising”; the researchers demonstrated a significant improvement in the properties of the potential fields algorithm commonly used in mobile robotics.