Print version

Advanced traffic management skills for autonomous vehicles

DSTO has participated in two research projects recently to give unmanned aerial vehicles (UAV) the capability of avoiding collisions in flight with their own kind, and to help unmanned ground vehicles (UGV) autonomously establish the best way to traverse difficult terrain

UAV is use during trails at Marulan test site

In 2004, DSTO entered into a collaboration with the Singaporean Defence Science Organisation (DSO) to develop and test the ability of an onboard collision avoidance system (OCAS) aimed at keeping multiple airborne UAVs safely apart.

This two-year program was conducted under DSTO’s Automation of the Battlespace Initiative and the Australia-Singapore agreement on Multi-UAV Collision Avoidance.

The impetus for research arose from concerns that the airspace in future operational arenas could well become very crowded with UAVs.

The trial phase of the research was carried out late last year at the Woomera Test Range where twelve UAVs - six actual and six virtual - were flown on a collision course.

The virtual UAVs used in the trial were generated by DSTO with its Force Level Electronic Warfare Simulation Environment (FLEWSE), while the actual UAVs were provided by Melbourne-based company, Aerosonde. Other private-sector companies providing support included Swordfish Computing and Consunet.

Setting a collision course

Giving background to the conduct of the exercise, DSTO researcher Andrew Bailey says, “The likelihood of collision was set to be intentionally high. The trial was designed to establish how well the UAVs could identify the risk of impact, share this information, and modify their flight paths to avoid impact.”

The process of attaining a collision course for the UAVs to avert, however, was not at all straightforward. Local weather conditions, with high temperatures that gave rise to strong thermals and variable wind patterns, made it difficult to accurately synchronise a single impact point for the actual UAVs used in the trial.

“The solution,” explains Bailey, “was to adjust the speed of the UAVs with our ground control stations to ensure the craft would reach the impact point all at the same time.”

Once the UAVs were on target, the ground controllers directed the UAVs to independently engage their OCAS to avoid collision, which they all readily managed.

The successful outcome of this trial, involving as it did twelve UAVs, is thought to be a world-first.

While some of the OCAS functions for the test were performed by ground-based equipment, research is under way to build these functions into the UAV craft so that in future they will be carried out entirely on board.

UAV provides pathfinding eyes for UGV

Another recent trial under DSTO’s Automation of the Battle Space Initiative was undertaken to study improvements in UGV performance made possible by integrating UAV and UGV operations.

The trial, involving the University of Sydney, the University of Adelaide and the Australian Defence Force Academy, was conducted at the University of Sydney’s Marulan test site in New South Wales.

A small helicopter UAV, carrying a miniaturised payload of electro-optic and laser radar sensors, flew over the terrain to produce a series of three-dimensional (3-D) strip images, which were fused into a single 3-D map of the area. These were then downloaded into the UGV to assist with its autonomous route planning processes.

The UGV, also equipped with navigation sensors including laser radar, Global Positioning System, Inertial Navigation System and electro-optic cameras, was capable of developing a 3-D view of its surrounds from a ground perspective.

Using the laser-generated maps from the UAV, the UGV was able to autonomously plan and then navigate its route through a bush setting. The 3-D aerial maps allowed the UGV to avoid obstacles such as trees, bushes and steep or uneven sections of terrain.

An impressively smart system

At the commencement of the trial, the UGV was instructed to travel between two points at extreme ends of the test site, separated by some thick scrub and rough terrain.

The on-board computers of the UGV then calculated the best route to take, with the unit remaining stationary throughout this process. When it started its journey, the course it embarked on came as a surprise to all present.

DSTO researcher Jean-Pierre Gibard recalls that “the UGV turned 180 degrees and headed for the dirt road through the test site which, though longer, was logically the best route.”

The trial established that the aerial mapping data from the UAV significantly improved the robustness and efficiency of UGV route planning and navigation processes. The 3-D UAV maps were especially useful for identifying sloping terrain and low-lying obstacles that could not be detected by the UGV’s sensors.

The success of this trial, hailed as an Australian first, is seen to be a significant advance in DSTO’s overall quest to develop a capability for unmanned vehicles and other robots to carry out dangerous tasks for the Australian Defence Force in place of personnel.

Back to the top