LTU-DENSO project aims to create smarter, safer autonomous cars

Giving undergraduates hands-on experience in advanced research while improving the “brains” of autonomous vehicles, letting them react to multiple sensor inputs faster and improving the safety of drivers, passengers, and pedestrians. That’s the aim of a two-year, nearly $200,000 research project involving several departments in the Lawrence Technological University’s College of Engineering. It’s funded by the DENSO North America Foundation (DNAF), the philanthropic arm of DENSO, a leading mobility supplier with North American headquarters nearby in Southfield.

Mostafa D. Rahbar, PhD, assistant professor in LTU’s Department of Electrical and Computer Engineering, is principal investigator on the project. Nabih Jaber, PhD, chair of electrical and computer engineering, and George Pappas, PhD, both ECE faculty, are co-principal investigators.

“My field of expertise is computer vision, applying computer vision for autonomous projects,” Rahbar said. “I worked on medical robotics in my PhD, and then, in my postdoc, the concept of sensor fusion, machine learning, real-time processing, and embedded systems in the field of autonomous vehicles, to address safety concerns.”

The multidisciplinary project will create hands-on training for students in LTU’s electrical and computer engineering, mechanical engineering, audio engineering technology, and architectural engineering programs, who will collaborate to measure roadside acoustic data in real time and implement control strategies to enhance vehicle safety. Students from the mathematics and computer science program in LTU’s College of Arts and Sciences will also be involved. All students will engage in hands-on labs and workshops, working on tasks such as sensor integration, data collection, and algorithm development.

The $192,197 grant funded the purchase of cameras, acoustic sensors, LiDAR (laser radar), and other equipment to create a car with multiple sensor inputs. The project will proceed in four phases:

• Sensor fusion and integration: Develop a framework to fuse data from all types of sensors, providing a more accurate and reliable perception of the vehicle’s surroundings
• Machine learning algorithms: Implement and train advanced machine-learning models for object detection, classification, and predictive path planning, using the fused sensor data to enhance decision-making processes in autonomous vehicles
• Real-time data processing: Design and optimize embedded systems capable of processing large volumes of sensor data in real time, ensuring timely responses to potential collision threats
• Field Testing and Validation: Conduct extensive field testing and validation of the developed system under various environmental conditions to ensure reliability and robustness