Advanced Driver Assistance Systems & Autonomous Driving

Feustel's research teams specialize in the development of advanced sensor fusion systems that integrate data from cameras, radar, lidar, and ultrasonic sensors to create comprehensive environmental understanding exceeding the capabilities of any individual sensing modality. This specialty addresses the critical challenge of achieving robust perception across diverse operating conditions by leveraging complementary sensor strengths while intelligently managing conflicting information and sensor degradation scenarios. Engineers develop sophisticated fusion algorithms utilizing deep learning architectures that operate at multiple abstraction levels, from low-level signal fusion through high-level semantic interpretation, enabling detection and tracking of vulnerable road users, vehicles, infrastructure elements, and dynamic obstacles with high reliability. The facility's testing infrastructure validates perception system performance across extensive scenario libraries incorporating varying lighting conditions, weather phenomena, and geometric configurations, ensuring that fused sensor systems maintain accurate environmental understanding when individual sensors experience reduced performance or temporary occlusion.

The facility maintains specialized capabilities in the creation and utilization of digital twin environments that provide high-fidelity virtual representations of real-world locations, enabling comprehensive autonomous system validation without the safety risks and logistical constraints of extensive on-road testing. This specialty encompasses the development of physics-based simulation platforms that accurately model sensor physics, vehicle dynamics, traffic behavior, and environmental conditions to create realistic test scenarios for perception, planning, and control algorithm evaluation. Engineers utilize generative artificial intelligence techniques to automatically create diverse scenario variations including rare edge cases, adversarial conditions, and complex multi-agent interactions that would require impractical amounts of real-world driving to encounter naturally. Feustel's unique capabilities include the integration of real sensor data with simulated environments, enabling software-in-the-loop and hardware-in-the-loop testing that validates system performance using actual production-intent computing hardware and sensor components within virtual scenarios that replicate challenging real-world situations.

Feustel has established dedicated expertise in the development of onboard computing architectures capable of executing sophisticated autonomous driving algorithms within strict latency, power consumption, and thermal constraints imposed by automotive deployment environments. This specialty focuses on optimizing neural network inference for specialized accelerators through quantization, pruning, and architecture search techniques that maintain perception accuracy while achieving real-time performance on energy-efficient hardware. Engineers develop custom computational pipelines that balance workload distribution between centralized processing units and distributed edge computing elements, ensuring that safety-critical functions maintain deterministic response times while maximizing the utilization of available computational resources. The facility's edge computing laboratory enables comprehensive characterization of processing system performance including inference latency under thermal stress, power consumption across various workload profiles, and fault tolerance mechanisms that ensure graceful degradation during hardware anomalies or software exceptions.

The facility specializes in the development of vehicle-to-everything communication systems that enable cooperative perception, intention sharing, and coordinated maneuvering between connected vehicles, infrastructure elements, and vulnerable road users. This specialty addresses the integration of dedicated short-range communication and cellular vehicle-to-everything technologies with onboard autonomous systems, creating architectures that leverage external information sources to extend perception range beyond onboard sensor limitations and improve decision-making through awareness of traffic signal timing, road hazard warnings, and surrounding vehicle intentions. Engineers develop sophisticated security protocols that authenticate communication sources, encrypt sensitive data exchanges, and detect anomalous messages that could indicate cyber attacks or system malfunctions. Feustel's testing infrastructure validates V2X system performance including communication range, message latency, and interoperability across diverse equipment implementations, ensuring that cooperative autonomy functions operate reliably in mixed-traffic environments where connected and non-connected vehicles share roadways.

Feustel maintains specialized capabilities in the creation, maintenance, and utilization of high-definition maps that provide centimeter-level localization accuracy and detailed semantic understanding of road infrastructure essential for safe autonomous navigation. This specialty encompasses the development of mapping systems that capture precise lane geometry, traffic control device locations, road surface characteristics, and dynamic layer information including construction zones and temporary traffic patterns. Engineers create robust localization algorithms that combine high-definition map data with real-time sensor observations including visual features, lidar point clouds, and satellite positioning to maintain accurate vehicle pose estimation in GPS-denied environments such as urban canyons, tunnels, and dense foliage. The facility's research addresses the critical challenge of map freshness and change detection, developing automated systems that identify infrastructure modifications through fleet-based data collection and crowd-sourced observation, ensuring that autonomous systems operate with current, accurate environmental representations that reflect real-world conditions.