Self-driving cars are displayed at the large motor shows, but Oskar Ljungqvist, newly promoted doctor at the Department of Electrical Engineering, discusses in his thesis another field: self-driving trucks with trailers. This combination will initially find application in mines, ports, and other closed environments.
Motion-planningHe has developed motion-planning and control algorithms that instruct the truck how it should move from its current position to another and end up with a desired configuration. This may involve both forward and backward motions, and it is of course important to keep track of the long trailer. Sensors and mathematical models take care of the latter task. Specially tailored controllers make sure that the computed motion plan is followed. The methods used and developed in the thesis work combine tools from automatic control, robotics, and optimisation.
Oskar Ljungqvist Photo credit Karl Öfverström“The manoeuvres should be as intuitive as possible. If it is possible to solve the task simply by driving forwards, the system will do so”, says Oskar Ljungqvist.
The planner uses a map of the surroundings in which various obstacles have been placed. New obstacles are detected by sensors on the vehicle. Based on this information, the planner calculates the best motion plan to reach the desired location.
“The planning algorithm starts with a number of precomputed basic manoeuvres, which it can automatically combine in different ways to solve the problem. The calculated motion plan can then be improved using methods from numerical optimisation”, says Oskar Ljungqvist.
LIDAR sensors mounted on the truck emit and detect laser light reflected from the trailer’s body. Mathematical models are then used to estimate distances and angles between the truck, dolly and trailer.
“There are relatively few control signals for a vehicle of this complexity. The angles between the truck and the trailer are particularly important – it’s important that the trailer doesn’t jack-knife,” says Oskar Ljungqvist.
LIDAR sensorsHe shows theoretically in the thesis that the vehicle follows the optimal motion plan it has computed, for both forwards and backwards motion, and he has developed new methods for verification. After practical tests in collaboration with Scania, he has been able to modify the system.
“The most recent type of controller is aware of other modules’ limitations, and can make sure that the truck is controlled such that the angles between the truck and the trailer can always be measured using the rear-view LIDAR sensor”, says Oskar Ljungqvist.
If external sensors are mounted on fixed items such as, for example, a loading bay, the system also operates in GPS-denied environments, such as underground.
He has also investigated the possibility of applying the results to the Skogsforsk concept ETT Modular System for Timber Transport. Here, a timber truck is given one more trailer, and its length increased from 24 to 30 meters. This increases the loading capacity without significantly increasing the fuel consumption. However, the additional trailer makes it much more difficult to manoeuvre in confined spaces, and reversing the vehicle is also a challenge. Oskar Ljungqvist’s system facilitates or completely automates these tasks.
It might be interesting in the future to redesign the vehicles themselves.
“The vehicles in use today have been designed for a human driver. Several vehicle manufacturers are working to optimise them for autonomous-use cases, where a person can be located at a remote facility and control several vehicles. An example is vehicles in an underground mine. The system may cope 99% of the time, but it might be necessary for a remote driver to intervene and take control in extreme situations”, says Oskar Ljungqvist.
Oskar Ljungqvist has worked on his doctoral thesis at the Division for Automatic Control, as one of a group of doctoral students working with motion planning, optimal control and real-time optimisation, led by Associate Professor Daniel Axehill. The work has been a project within the Strategic Vehicle Research and Innovation (FFI) programme, in collaboration with Scania CV. Oskar Ljungqvist has also been an affiliated doctoral student within WASP, the Wallenberg AI Autonomous Systems and Software Program.
Motion Planning and feedback control techniques with applications to long tractor-trailer vehicles, Oskar Ljungqvist, Division for Automatic Control, Department of Electrical Engineering, Linköping University 2020.
Translated by George Farrants