Highly Automated Driving on Highways Based on Legal Safety

Abstract

Vehicle automation is proposed as one of the solutions to make transport safer, more comfortable and more environmentally friendly. It is gradually being introduced through Advanced Driver Assistance Systems (ADAS). ADAS started as Anti-lock Braking Systems (ABS) and have now extended to Adaptive Cruise Control (ACC) and Lane Keeping Assist Systems (LKAS). This work aims to contribute to this evolution, by discussing how driving systems can share the road with human drivers. It presents the legal safety concept for the design of a highly automated driving system for highways. The legal safety concept proposes to base driving system design on traffic rules. This allows fully automated driving in traffic with human drivers, without necessarily changing equipment on other vehicles or infrastructure. The driving system uses traffic rules to predict legal or nonlegal trajectories of objects in its perception zone and worst-case objects outside its perception zone. If all objects respect traffic rules, accidents will be avoided. If not, driving defensively will avoid most accidents. Today, international law allows highly automated driving, but not yet fully automated driving. The driving system interacts with the human driver, via human rules. The HAVEit project (European Seventh Framework Programme) and ABV project (French National Research Agency) propose human rules based on the horse-rider metaphor (Hmetaphor). If needed, the driving system takes over control in order to avoid accidents. If the human driver cannot continue driving, the driving system brings the vehicle to a safe standstill on the emergency lane. The consequences of these human rules on driving system design are explored. System rules form the third set of rules of the legal safety concept. With system rules, system components respect the limitations of other system components. The requirements on perception, control and Human-Machine Interface (HMI) components of the legal safety system are discussed. The decision component, which is the central component of the legal safety system, is completely worked out from requirements to design. The legal safety system has been implemented on PC and automotive Electronic Control Units (ECUs). The integration and validation of legal safety components on LIVIC, HAVEit and ABV demonstrators are presented. The work concludes that, for highway environments, legal safety decision, control and HMI can be achieved with state-of-the-art technology, and legal safety perception could be available in medium term.

SPOT: A tool for set-based prediction of traffic participants

“…This, however, is required to ensure safety as described in [8] (referred to as zone model) or in [9]. Our proposed tool SPOT uses reachability analysis to predict the occupancy of surrounding traffic participants in a rigorous and set-based way.…”

Set-Based Prediction of Traffic Participants on Arbitrary Road Networks

“…d) Uncountable set of future behaviors: In the architecture proposed in [33], the importance of set-based prediction of other vehicles is highlighted, but no algorithm to formally compute the occupancy prediction is proposed. In our previous work [3], we utilize techniques from reachability analysis to rigorously compute an over-approximation of the occupancy of surrounding vehicles.…”

Formalising Traffic Rules for Accountability of Autonomous Vehicles

“…Vanholme et al [19] have previously elicited, simplified, and paraphrased the traffic rules from the Vienna Convention on Road Traffic for highways. The list of traffic rules considered in this work is the subset of rules from their work and it is shown in Tab.…”