A Formal Approach for the Design of a Dependable Perception System for Autonomous Vehicles

Abstract

The deployment of autonomous vehicles is contingent on trust in their ability to operate safely. However, the assurance that they can accommodate failures and changing weather conditions to maintain limited functionality requires the development of rigorous design and analysis tools. This paper presents a formal approach for the design of multisensor data fusion systems that support adaptive graceful degradation through the smart use of sensor modalities. A coloured probabilistic time Petri net is used to model known algorithms in a multi-sensor fusion scheme. The specification of safety requirements in terms of confidence levels conditions the outcome of the reachability analysis. The characteristics of a credible solution are then provided to the embedded safety module as support for online reconfiguration and decision making tasks. The validity of the approach is illustrated through an example outlining the capabilities of currently available perception systems, for the purpose of deploying autonomous vehicles on public roads.

Set-Based Prediction of Traffic Participants Considering Occlusions and Traffic Rules

“…We solely focus on motion prediction of other traffic participants [24]- [26], which is an integral part of motion planning [27]- [29] and risk assessment [24], [30]. The following related aspects are beyond the scope of this paper: extracting the information of surrounding traffic participants from sensor measurements [31]- [33], the uncertainty of these measurements [34]- [36], and implications on the prediction for connected vehicles [37], [38].…”

Methodology and Architecture for Safety Management

“…Self-awareness on system's abilities and limitations: In the matter of monitoring, Reschka et al [7] reports the needs to provide a permanent online monitoring of vehicle capabilities, as well an adequate modelling tool to support appropriate and safe decisions. Accordingly, the framework perceives the current performance of the system during operation, by knowing the range of actions of the system and its limitations.…”