In this paper we consider a fleet of self-driving cars operating in a road network governed by rules of the road, such as the Vienna Convention on Road Traffic, providing rides to customers to serve their demands with desired deadlines. We focus on the associated motion planning problem that tradesoff the demands' delays and level of violation of the rules of the road to achieve social optimum among the vehicles. Due to operating in the same environment, the interaction between the cars must be taken into account, and can induce further delays. We propose an integrated route and motion planning approach that achieves scalability with respect to the number of cars by resolving potential collision situations locally within so-called bubble spaces enclosing the conflict. The algorithms leverage the road geometries, and perform joint planning only for lead vehicles in the conflict and use queue scheduling for the remaining cars. Furthermore, a framework for storing previously resolved conflict situations is proposed, which can be use for quick querying of joint motion plans. We show the mobility-on-demand setup and effectiveness of the proposed approach in simulated case studies involving up to 10 selfdriving vehicles.

While motion planning under temporal logic specifications has been addressed in several state-of-the-art works, spatial aspects have been so far largely neglected. In this work, we enrich the semantics of robot motion specifications by including preferences on spatial relations between its trajectory and various elements in its environment. The spatial preferences are given in a fragment of Signal Temporal Logic (STL) on top of complex missions in syntactically co-safe Linear Temporal Logic (scLTL). We propose a cost function with user-specified parameters, which determines the compromise between efficiency and spatial robustness of a trajectory.  The proposed modification of the incremental sampling-based RRT$^\star$ driven by this cost function guarantees that the motion plan (if found) simultaneously satisfies the mission and asymptotically minimize the cost. The paper includes several case studies showcasing the effects of the user-adjustable parameters on the resulting trajectories.

We present an approach for intention-aware motion planning in an autonomous driving scenario, where a vehicle aims to traverse a road segment as quickly as possible, while constrained by road rules encoded in syntactically co-safe linear temporal logic. We show that by combining the RRTx algorithm with trajectory prediction using Mixed Observable Markov Decision Processes (MOMDP), we can achieve least-violating behavior wrt. mission completion time and the road rules, while ensuring that the likelihood of collisions remains below a user specified threshold. We illustrate the validity of our approach using simulations of a variety of traffic scenarios. 

In this work, we combine a sampling based motion planner with path non-existence verification. We show that, using our approach, it is possible to: 1) provide termination criteria to sampling-based motion planners, based on the problem specific constraints; 2) generate a method for rejection sampling that can adapt to the scenario, for instance, leveraging the relative priority between constraints. Furthermore, we describe a language-guided sampling technique based on IRRT?, that utilizes the results from the path non-existence verification procedure to reduce runtime of the underlying motion planner. The approach is studied using route and motion planning for an autonomous vehicle that aims to service transportation tasks while constrained by rules of the road. We evaluate the proposed approach using a set of real-life inspired traffic scenarios. Finally, we show that the resulting runtime of a motion planner with path non-existence verification is significantly shortened when compared to the same motion planner with traditional termination criteria, such as a resource or time limit.

Driving styles play a major role in the acceptance and use of autonomous vehicles. Yet, existing motion planning techniques can often only incorporate simple driving styles that are modeled by the developers of the planner and not tailored to the passenger. We present a new approach to encode human driving styles through the use of signal temporal logic and its robustness metrics. Specifically, we use a penalty structure that can be used in many motion planning frameworks, and calibrate its parameters to model different automated driving styles. We combine this penalty structure with a set of signal temporal logic formula, based on the Responsibility-Sensitive Safety model, to generate trajectories that we expected to correlate with three different driving styles: aggressive, neutral, and defensive. An online study showed that people perceived different parameterizations of the motion planner as unique driving styles, and that most people tend to prefer a more defensive automated driving style, which correlated to their self-reported driving style.

To display perceivably different driving styles is an important ability for autonomous vehicles in real-life traffic scenarios. By encouraging predictable driving styles, we can promote trust and collaboration between vehicle and human drivers in traffic. However, many motion planners lack the ability to provide different driving styles using one method.As a result, many applications are overly defensive to ensure safety, and can not be tailored to the users' preferences. In this work, we build on our previous works on encoding perceivable driving styles using Signal Temporal Logic (STL) and generating motion planning trajectories using user preferences. We refine our previously proposed spatial constraints based on the Responsibility-Sensitive Safety (RSS) model. We illustrate how the motion planner can be parameterized to produce aggressive, neutral and defensive driving behaviors, respectively.  We evaluate the resulting driving styles on a set of real-life inspired driving scenarios, modeled in the Carla simulator and provide a detailed statistical analysis of the generated trajectories.