Software Issues in Critical Transportation Systems

Principal Investigator:

Mats Heimdahl, Associate Professor, Computer Science and Engineering


Project Summary:

Computer software plays an increasingly important role in safety-critical computer controlled systems; systems where incorrect operation of the software could lead to loss of life, or substantial material or environmental damage. Systems of particular interest in this proposal are transportation related systems such as autonomous vehicle control, automated highway systems, driver information systems, driver fatigue detection, and the virtual bumper. Currently, the ability to develop high quality software for such systems is limited. Software is often poorly engineered and software failures are common. Failures linked to software can be classified in three general categories 1. the software is simply faulty and does not perform its intended function 2. sensors or actuators in the system fail and the software does not detect the failure and thus, performs erroneous operations, and 3. sensors or actuators fail, the software detects the failure, but the software takes erroneous recovery actions. Many ITS systems must provide a certain level of service in the face of disturbances and failures: system shutdown is not an acceptable solution. For example, a vehicle in a platoon or under autonomous control must provide safe means of recovering from failures, anything less would be unacceptable from ethical, regulatory, and legal standpoints. This proposal deals with the application of formal design and verification methods in the area of critical transportation related software. In particular, it brings software engineering expertise to the design and analysis of transportation related software. Software is often engineered poorly and cannot effectively handle sensor failures. We plan to study traffic control software that uses loop detectors or cameras, vehicle following software, pedestrian control software, and other real-time transportation related software systems. The approach has a solid foundation in software-safety and we plan to focus on the critical issue


Project Details: