You are here

Progress with urban autonomous vehicle flight

9 September 2019
Defence scientist Ricardo Cannizzaro.
Defence scientist Ricardo Cannizzaro.

DST robotics specialist Ricardo Cannizzaro is using a unique indoor flight lab to integrate and test new algorithms, hardware and sensors on small uninhabited aircraft systems (UAS).

“It’s a great environment to test things and demonstrate proofs of concept,” Cannizzaro explains. “If an experiment doesn’t go entirely to plan, the drone can land on the soft mats and is contained by the netting. Our motion capture system gives us position and orientation information of aircraft as they fly, allowing us to compare the outputs of various position estimation algorithms against ground truth. Because GPS signals are not available inside, we can also pipe that data in real time to the aircraft’s flight controller and tell it ‘this is where you are to close the loop on its positioning system.”

A robot’s ability to build a map of its surroundings and calculate its own position, known as simultaneous localisation and mapping (SLAM), is a common robotics problem that research groups around the world are tackling. The approach taken by this team, led by Dr Jennifer Palmer, is for the UAS to use a 2D scanning laser sensor, monitoring changes between scans as it moves to build a picture and thus a map, which it uses to ‘find’ itself in.

Rising to the challenge on CUE

For The Technical Cooperation Program (TTCP) Contested Urban Environment (CUE) 2018 strategic challenge in Canada last year the team deployed a 2D SLAM algorithm, tweaking it for the specific application.

“And for our workhorse quadcopter, we’ve developed software that handles mission management. It maintains a list of waypoints (locations to which the aircraft flies), makes decisions to plan a safe path around obstacles and issues take-off commands,” explains Cannizzaro.

The team has constructed the quadcopter from low-cost hobbyist components in a bid to see the performance that can be achieved at this price-point. A new rapid prototyping capability is proving very useful – allowing for the custom design and build of 3D components to quickly modify the quadcopter when, for example, new sensors need to be mounted.

At the recent CUE 2019 challenge in the US, several improvements in UAS autonomy were successfully demonstrated. “We had the incredible opportunity to demonstrate our UAS capability in three urban environments in New York City: in a subway tunnel, a high-rise building, and a downtown streetscape. We learned more about our system and recorded a lot of valuable data, which was also of great interest to the international community.”

For a mechatronics and robotics engineer Cannizzaro says his work is a really exciting combination of electronics, mechanics and software engineering.