StreetWise is a data-driven methodology developed in TNO’s program to make traffic more efficient, sustainable, and safer. It offers a solution to “knowing” foreseeable traffic situations that trucks will encounter on a trip from one hub to another. The methodology provides real-world scenarios and test cases for the development and assessment of Level 4 driving systems through automated analysis of driving data. Torc is implementing StreetWise as a tool in its validation pipeline to analyze its real-world drive data. From the database parameters, substantial statistics are derived, keeping the original scenarios classified. The resulting scenario information enables Torc to apply thorough, virtual validation, according to state-of-the-art international safety assessment processes.
“Torc has 17 years of experience in pioneering safety-critical self-driving applications, and our technology is designed with safety as the priority. We take the driving behavior of the safest, most experienced drivers on the road as our role model,” said Axel Gern, senior vice president of engineering and managing director for Torc Europe. “The StreetWise methodology offers the foundation for safety assessment. In addition, our shared beliefs regarding traffic safety ensure that TNO understands our needs and can provide the support to implement and scale up this methodology.”
Collective traffic scenarios
For safety assessments, the industry is using parametrized traffic situations as defined in the “Pegasus Project.” The core of the StreetWise method is a database in which all parameterized traffic scenarios are stored. The scenarios have been identified from driving data collected from various fleets of data collection vehicles. Modeling scenarios in a parameterized way in computer-readable format provides the necessary input for smart simulation software and a means to calculate the residual risk of autonomous driving. TNO has developed its methodology such that it even allows for building a collective scenario database among specific industrial partners who make their data available for this purpose, without exchanging sensitive driving data between the partners. This permits partners to scale up to a scenario catalogue that is complete for different cities, countries, and continents, without the need to run their own expensive testing or data collection campaigns in each of the regions.
Feasible end-to-end safety assessment
The TNO and Torc collaboration includes the implementation and upscaling of TNO’s StreetWise software in TORC’s fleet operation. Torc will use the software to process fleet records daily and provide development engineers with traffic understanding. “Autonomous transport plays an essential role in TNO’s mission towards more efficient, sustainable, and safer traffic and transport. The new safety challenges require legislation and standardization within the industry. We are very pleased that Torc will be the first to operationalize our methodology, bringing adoption and standardization in the field of safety validation one step closer,” said Olaf Op den Camp, safety assessment specialist, TNO.