Situation Understanding & Motion Planning
Autopia has a solid background on motion planning for different types of structured environments, considering (I) multi-sensor perception, mapping and localization uncertainty under a common framework relying in dynamic occupancy grids, and (II) human-comfort-acceptable motion primitives. To assess the behavior of these strategies in a real car, maps and sensor-based world modelling needs to be developed, and AUTOPIA has started recently a research line in this direction, always considering uncertainty and low-fidelity maps.
Keywords: Sensor fusion, world modelling, path planning under uncertainty, speed planning
Decision Making & Control
One of the fields where AUTOPIA has traditionally focused its efforts on is guidance and control, testing both in simulation and on real vehicles a large panoply of model-based and model-free strategies, always looking for a robust behavior in the broader set of operational domains. On top of control and motion planning, a behavior planning is needed to cope with the large number of driving scenarios we face in urban environments. AUTOPIA is developing a novel approach based on stochastic reachability to predict the motion of the actors in the driving scene, and plan accordingly a behavior that meets an explainable trade-off between safety, comfort and usefulness.
Keywords: Reachability analysis, risk assessment, motion prediction, behavior planning, model-based and model-free control
Interactive & Cooperative driving
The ability of the automated vehicle to interact with other vehicles, VRU’s or passengers is one of the foundational interests of AUTOPIA. As a result, the group has a solid experience in safety-related applications with V2X communication (platooning, automated merging, overtaking, enhanced/cooperative perception). In addition to that, a key aspect we are recently researching is the adoption of the safest and more convenient SAE Level for each the driving situation and the driver attentiveness level, for which innovative HMIs are also developed. In this connection, a specific solution we are experiencing with in some projects is the use of tele-operation as a fallback.
Keywords: V2X communications, cooperative maneuvers, V2I interaction, HMI, tele-operation
Simulation, Integration & hands-on Experimentation
One of the unique selling points of AUTOPIA is the high number of vehicles that have been automatized from scratch, transforming commercial cars into operational automated vehicles (AV) with which novel standalone and cooperative solutions can be tested. To facilitate the cooperation with other entities that develop complimentary components, modular HW/SW architectures have been adopted relying on different middlewares (LCM, ROS). Autopia has a proven experience in implementing drive by wire systems, integrating a wide range of sensors and tuning microcontrollers and embedded high-performance computing devices.Since testing with a fleet of automated vehicles is far from being cost-effective, AUTOPIA has also develop co-simulation tools and work on a daily basis with a SiL and HiL advanced simulation framework for AVs.
Keywords: SiL and HiL simulation, drive by wire, embedded high performance computing, middleware-based integration, fail-operational systems