Bosch launches Project 3F to develop fault-tolerant autonomous vehicles that can still move safely even when they fail

According to foreign media reports, German Tier 1 supplier Bosch is working on a project called Project 3F - "Fault-tolerant driverless vehicles in the low-speed driving range", using driverless shuttles to transport tourists from tram stations to exhibition centers, or move containers full of packages in logistics centers. The main purpose of the autonomous shuttle is to be able to get from point A to point B safely, so the project focuses on preventing fault operations, that is, developing solutions to ensure that the autonomous shuttle can move safely even if there is a technical failure or an obstacle suddenly appears.

Project 3F, which received €4.3 million in funding from the German Federal Ministry of Economic Affairs, is led by Bosch and includes three other companies, a university and an institute: StreetScooter, RA Consulting, FZI Research Center for Information Technology, Finepower and RWTH Aachen University. The project team is focused on ensuring that even if a malfunction occurs, the vehicle does not fail completely and can continue to drive.

“Unlike highly automated passenger cars, driverless shuttles have different requirements to meet,” said Thomas Schamm, project coordinator at Bosch. “To operate without a safety driver, the shuttle needs to be able to monitor its systems autonomously, i.e. perform diagnostic tasks, and handle any technical faults found so that it can continue driving. At the same time, it also needs to be able to protect the system in the event of a major fault, such as stopping itself.”

One solution proposed by the project partners is to add redundant systems, i.e. to duplicate safety-related functions. For example, the researchers developed redundant systems for power supplies, so that the electric drivetrain and vehicle electrification systems are reliably protected. In addition, the researchers adapted and improved the sensor technology to the vehicle design. In order to reliably detect obstacles, the researchers installed several lidar and radar sensors at multiple locations around the vehicle, allowing it to detect the surroundings from different positions. By providing a 360-degree bird's-eye view and avoiding blind spots, a 3D protection zone is created, allowing it to detect not only obstacles on the road, but also objects such as hanging branches.

Another solution is to build fault tolerance, where the failure of a subsystem can be at least partially compensated by other functions. Bosch says this is a bit like the following situation: if the lights in the room suddenly go out, humans can still use other senses to feel around instead of completely failing. The shuttle behaves similarly. If it cannot see an area because leaves are stuck to the sensors or large objects such as trash cans in a certain direction completely block the line of sight, the car will slow down or omit the route that can no longer be detected.

In addition, the project is also working to ensure that the shuttle can react to changes on the established route. When any moving objects approach, the car is able to slow down, and if in doubt, it will keep its distance from unknown objects. On the other hand, when the vehicle recognizes familiar landmarks such as street lights, it will continue to move forward at full speed. If there are any urgent dangerous situations, the shuttle will also be able to take preventive measures and stop. Therefore, the goal of the project is to enable the autonomous shuttle to adjust its driving behavior according to the situation in real time, while continuing to drive automatically as much as possible, even if the system fails or encounters obstacles on the road.

The trip data and current technical status of the autonomous shuttle can also be transmitted to and from the vehicle, so information on three different functions, diagnosis, monitoring and control, can be transmitted back and forth, making it three times more powerful than telemetry, laying the foundation for a fleet of autonomous shuttles that can be remotely monitored, repaired, and even controlled (such as opening doors). This means that if a vehicle reaches the limits of fault detection and functional compensation, it can be helped, or if such vehicles only need simple regular maintenance.

Bosch said that the solution developed by the project is not only suitable for driverless shuttles, but can also provide strong support for the logistics industry. Project members have developed an auxiliary system for driver-vehicle interaction, which can accurately position the detachable body crane (a special vehicle used to move containers in logistics centers). The system is designed to move the vehicle under the gantry crane with centimeter-level accuracy, allowing the container to move quickly, but this requires precise positioning and the ability to automatically stop under the gantry. In practice, this type of automated operation can collect and position containers without error.

The above R&D projects were tested on several test tracks: at Bosch's Renningen Research Park, two shuttle buses tested traffic services near a site with pedestrians; at the Innovation Park near RWTH Aachen University and near a warehouse of Deutsche Post/DHL, a logistics vehicle was deployed to test the interaction between drivers and automated vehicles. (All pictures in the article are from autocarpro.in)