Search

We argue that object detectors in the safety critical domain should prioritize detection of objects that are most likely to interfere with the actions of the autonomous actor. Especially, this applies to objects that can impact the actor’s safety and reliability. To quantify the impact of object (mis)detection on safety and reliability in the context of autonomous driving, we propose new object detection measures that reward the correct identification of objects that are most dangerous and most likely to affect driving decisions. To achieve this, we build an object criticality model to reward the detection of the objects based on proximity, orientation, and relative velocity with respect to the subject vehicle. Then, we apply our model on the recent autonomous driving dataset nuScenes, and we compare nine object detectors. Results show that, in several settings, object detectors that perform best according to the nuScenes ranking are not the preferable ones when the focus is shifted on safety and reliability.

Coding conventions are a means to improve the reliability of software systems, and they are especially useful to avoid the introduction of known bugs or security flaws. However, coding rules typically come in the form of text written in natural language, which makes them hard to manage and to enforce. Following the model-driven engineering principles, in this paper we propose an approach for the management and enforcement of coding conventions using structured models. We define the Coding Conventions Specification Language (CCSL), a language to define coding rules as structured specifications, from which checkers are derived automatically by code generation. To evaluate our approach, we run a thorough experiment on 8 real open-source projects and 77 coding rules for the Java language, comparing the violations identified by our checkers with those reported by the PMD static analysis tool. The obtained results are promising and confirm the feasibility of the approach. The experiment also revealed that textual coding rules rarely document all the necessary information to write a reliable checker.

In the aftermath of a large-scale emergency, Unmanned Aerial Vehicles (UAVs) can play a key role as mobile communication systems supporting rescue operations on the ground. At the same time, the deployment of autonomous UAV swarms still poses severe challenges in terms of distributed mobility, swarm connectivity and mesh networking. To this purpose, we propose ELAPSE (aErial LocAl Positioning System for Emergency), a novel, distributed framework for aerial mesh deployment that supports discovery and multi-hop connectivity among rescue personnel and emergency requesters. ELAPSE integrates components of swarm mobility, positioning and Quality-of-Service (QoS) support, while targeting UAV devices at different levels of hardware complexity. Three contributions are provided in this study. First, we present a novel, bio-inspired swarm mobility algorithm which natively addresses QoS-based aerial mesh connectivity, coverage of the ground nodes and UAV collision avoidance through the abstraction of virtual springs. Second, we investigates its implementation when geo-location capabilities are not available: to this aim, we propose local-based and cooperative-based techniques through which each UAV can estimate the position of its neighbours, and hence correctly adjust its direction and speed. Third, we analyze the feasibility of the ELAPSE framework through a twofold evaluation: i.e. a large-scale OMNeT++ simulation showing the effectiveness of the distributed mesh formation and localization techniques, and a small-case ground robotic testbed demonstrating the impact of QoS mechanisms on the system operations.