Ibeo develops automated turnkey solution for ADAS and autonomous driving sensor validation

According to foreign media reports, automotive LiDAR supplier Ibeo has developed a new automated overall solution for ADAS and autonomous driving sensor verification. The ibeo.Reference tool chain automates manual annotation, saving a lot of time and cost while ensuring high performance, and can be used as a reference for almost all traditional sensor systems. The ibeo.Reference system can meet the high demands of OEMs and help Tier 1 build more powerful systems.

(Image source: Ibeo)

ADAS and automated driving systems must undergo complex validation before installation in vehicles and before series production. Regardless of the technology used in the sensor system (camera, radar, or lidar), each system requires reference data that must be compared with the data obtained during testing. Ibeo's recording system is installed in the vehicle together with the manufacturer's prototype system and records the data simultaneously. In addition, the recording system can be used to create scenarios for the development of ADAS or AD algorithms.

The recorded data collected during validation drives can amount to thousands of hours. Every hour of test driving requires manual evaluation, resulting in up to 300 hours of annotation work, which is a tedious and expensive task that may be inaccurate in various aspects.

"Manually checking this data is currently very time-consuming and costly in many new developments," said Philip Neitzel, Product Manager at ibeo.Reference. To sort through the large amounts of data recorded, the industry often outsources services. In many cases, hundreds of employees manually assess every meter of road to determine whether the sensor being tested detects a vehicle, person, lane marking or guardrail, etc., so that the data remains consistent.

The Ibeo Validation Bundle in Ibeo’s new tool chain solves this problem. “As part of the ibeo Validation Bundle, the ibeo Auto Annotation Software provides an automated turnkey solution that algorithmically generates reference data from input data. If required, this data can be manually processed to generate ground truth data, i.e. data that truly represents the vehicle environment,” Neitzel explained.

With the new ibeo.Reference toolchain, users receive all the elements required for sensor validation from a single source:

• The system of record consists of hardware and software for operating in reference data and now also supports selected third-party lidar sensors for the first time

  
  

• ibeo Auto Annotation is a scalable, cloud-compatible software for automatically creating annotated reference data

  
  

• The editor further processes the reference data to generate ground truth data

Neitzel added: “It is very important for us to offer complete systems that cover almost all challenges that our customers face in combining lidar components and to regularly expand our technology. In this way, we enable our customers to develop faster, more accurately and at lower costs, and thus get closer to the vision of fully autonomous mobility.”