Autonomy and Vehicle Control at Vehicular Systems

Autonomous traffic
Autonomous traffic iStock/AkaratPhasura

Autonomous cars and automated transport has the potential to increase efficiency and safety. Our research aims at methods for realizing resiliant and robust automation of vehicles under significant uncertainties in sensing and surrounding vehicles. Key topics in our research are planning and robust vehicle control, optimization of avoidance manuevers, and modelling of surrounding agents behavior in traffic.

Connected and Automated Transport (CAT) systems have the potential to increase efficiency, both for passengers of the vehicle and for other traffic participants. Therefore, a main objective of our research is to develop methods and techniques for fault-tolerant decision-making and resilient behavioral control of autonomous ground vehicles in challenging, possibly multi-vehicle, traffic situations.

It has been shown (Olofsson, B. et al., 2020) how automated and autonomous functions in a vehicle has the potential to significantly increase safety and reduce the number of fatal and severe accidents in traffic. The research considers a range of vehicles, from passenger cars to heavy-duty trucks.

Currently, our research activities fall into the three main categories of;

  • planning and control
  • modeling and prediction of surrounding traffic
  • optimal vehicle maneuvers

Planning and Control

The task of motion planning and control for autonomous vehicles is challenging for many reasons, and one significant difficulty is the interaction with surrounding vehicles and the many uncertainties affecting the vehicle motion and environment. Perception and processing of sensor data is an important part of an autonomous vehicle to create situational awareness.

Our research focuses on leveraging on such awareness to provide reliable autonomy by developing methods and techniques for planning, control, and prediction of the behavior of the surrounding environment. Examples of key research results utilizing different techniques to safely plan and navigate in dynamic environments are (Morsali, M. et al., 2020 and Mohseni, F. et al., 2020). These methods have been further developed in, e.g., (Zhou, J. et al., 2022), to explicitly predict and handle the interaction between surrounding vehicles when planning the motion of an autonomous vehicle in an MPC framework.

Uncertain behavior of surrounding traffic can also be handled by anticipating possible dangerous situations that can occur, and then incorporating such information into the motion planner and controller (Fors, V. et al., 2022). The video below illustrates how such control can be applied in decision-making in challenging and uncertain traffic environments.



Modeling and Prediction of Surrounding Traffic

Picture shows vehicle in traffic Another active area of research in the group is the development of advanced models for the behavior of the surrounding traffic. For example, one such traffic situation is to predict if neighboring vehicles will remain in their lane or change left or right, as illustrated by the figure to the right. The paper (Westny, T. et al., 2021) develops neural-network models for predicting the intention of surrounding traffic in multi-vehicle situations on highways.

Optimal Vehicle Maneuvers

Vehicular systems Photo credit Toa55 The research on optimal vehicle maneuvers in the group develops methods for motion planning and control at the limit of vehicle capacity. The aim is to enable autonomous handling of time and safety-critical situations, such as when an unexpected obstacle appears on the road or when driving too fast when entering a low-friction road section. In such situations, it is desirable to utilize the maximum of the available friction between the vehicle tires and the road.

The research has resulted in both methods for formulating and solving optimal vehicle-maneuvering problem using dynamic optimization (e.g, Anistratov, P. et al., 2022), and methods for real-time control of vehicles in safety-critical situations using a force-centric perspective (e.g., Fors, V. et al., 2020).

Key References

  • V. Fors, B. Olofsson and E. Frisk, “Resilient Branching MPC for Multi-Vehicle Traffic Scenarios Using Adversarial Disturbance Sequences” in IEEE Transactions on Intelligent Vehicles, doi: 10.1109/TIV.2022.3168772.
  • M. Morsali, E. Frisk and J. Åslund, “Spatio-Temporal Planning in Multi-Vehicle Scenarios for Autonomous Vehicle Using Support Vector Machines” in IEEE Transactions on Intelligent Vehicles, vol. 6, no. 4, pp. 611-621, Dec. 2021, doi: 10.1109/TIV.2020.3042087.
  • F. Mohseni, E. Frisk and L. Nielsen, “Distributed Cooperative MPC for Autonomous Driving in Different Traffic Scenarios” in IEEE Transactions on Intelligent Vehicles, vol. 6, no. 2, pp. 299-309, June 2021, doi: 10.1109/TIV.2020.3025484.
  • V. Fors, B. Olofsson and L. Nielsen, “Autonomous Wary Collision Avoidance,” in IEEE Transactions on Intelligent Vehicles, vol. 6, no. 2, pp. 353-365, June 2021, doi: 10.1109/TIV.2020.3029853.
  • P. Anistratov, B. Olofsson and L. Nielsen, “Analysis and design of recovery behaviour of autonomous-vehicle avoidance manoeuvres” in Vehicle System Dynamics, vol. 60, no. 7, pp. 2231-2254, 2022., doi: 10.1080/00423114.2021.1900577
  • T. Westny, E. Frisk and B. Olofsson, “Vehicle Behavior Prediction and Generalization Using Imbalanced Learning Techniques”, 2021 IEEE International Intelligent Transportation Systems Conference (ITSC), 2021, pp. 2003-2010, doi: 10.1109/ITSC48978.2021.9564948
  • J. Zhou, B. Olofsson and E. Frisk, “Interaction-Aware Moving Target Model Predictive Control for Autonomous Vehicles Motion Planning” in European Control Conference (ECC’22), pp. x-y, 2022, doi: 10.23919/ECC55457.2022.9838002
  • B. Olofsson, and L. Nielsen (2020). “Using crash databases to predict effectiveness of new autonomous vehicle maneuvers for lane-departure injury reduction”. IEEE Transactions on Intelligent Transportation Systems, 22(6), 3479-3490, doi: 10.1109/TITS.2020.2983553


Recent Doctoral Theses

About the Division

About the Department