“As major automakers accelerate the adoption of ADAS and autonomous driving functions in their cars, the automotive industry is moving in this direction. There are more and more test drives of fully automatic passenger cars. This “application note” discusses the LiDAR technology used by autonomous cruise vehicles and provides cost-effective solutions to key safety issues.
As major automakers accelerate the adoption of ADAS and autonomous driving functions in their cars, the automotive industry is moving in this direction. There are more and more test drives of fully automatic passenger cars. This “application note” discusses the LiDAR technology used by autonomous cruise vehicles and provides cost-effective solutions to key safety issues.
The core of a typical fully automatic cruise vehicle relies on two important central processing units (CPU)-the main CPU of the navigation system and the safety CPU, which work together to prevent collisions and ensure safety.
The main CPU of the navigation system can interact with one or more LiDAR sensors to provide large-scale and high-resolution positioning, mapping and collision avoidance. At present, many autonomous cruise vehicles rely on mechanical scanning LiDAR sensors, which are located on the top of the cruise vehicle and can perform remote detection and can achieve 360-degree coverage around the vehicle.
The radius of the blind spot area is affected by many factors, including the vertical field of view (FoV) of the mechanically scanned LiDAR sensor, the location of the sensor on the vehicle, and the size and shape of the cruiser. However, it is worth noting that mechanical scanning sensors are usually not specifically configured to obtain close-range data, or specifically select the best location to achieve full 360-degree coverage, which further leads to an increase in the blind spot area around the cruiser.
The figure mechanically scans the blind spots generated by LiDAR around the vehicle.
The ability to detect obstacles in the blind spot area is essential to ensure safe and successful autonomous driving. When the vehicle is turned on, if the surrounding environment is not detected and it is confirmed that there are no obstacles nearby, the cruise vehicle cannot start. In addition, during the operation of the cruising vehicle, for situations where it is about to be restarted after being stopped, the system needs to verify in advance whether people, vehicles or any other objects are approaching the cruising vehicle. For example, when picking up passengers, the system needs to confirm whether the passengers have boarded the bus or are still staying in the blind zone. Other objects in the blind area of the cruiser (such as convertibles, forklifts, trailers, etc.) also need to perform the same detection process.
Scanning the blind spot can have disastrous consequences, just like the child in the blind spot of the vehicle in this example.
When a dangerous situation is detected, the safety CPU prevents the cruising vehicle from moving or stops it completely. As an indispensable part of the latest automatic cruiser design, the safety CPU needs to perform short-to-medium-range collision avoidance detection so that the vehicle can be completely stopped when the main navigation system fails to identify obstacles or malfunctions.
The safety CPU is a low-power, cost-effective redundant system with low data rate requirements-but it must ensure complete coverage of its surrounding environment (360 degrees, full horizontal and vertical coverage). This poses a huge challenge because the requirement itself violates the basic concept of scanning-greater coverage usually means more data. For example, LiDAR scanning equipment returns 600,000 to 2.2 million points per second, depending on the resolution of the sensor and the number of vertical lines (16 to 64).
Find the right technology
In order to be able to cover the blind spots around the cruiser, we considered and tested different detection technologies (solutions such as ultrasonic, radar, and camera), but each technology has its limitations:
• Ultrasound: limited range, low resolution
• Radar: Detection of permeable surfaces and static objects
• Camera: lack of distance information and important weather gradient information
The standards adopted by other solutions are equally important. First of all, because the solution will play the most important role in the safety CPU, too much data should not be sent to the system to avoid data interface problems caused by data overload. In order to achieve 360-degree coverage, we estimate that the optimal data input volume for the safety CPU is 3,000 to 60,000 points per second.
The technology must also provide high enough resolution to detect small objects, and can achieve precise trajectories and speeds through tracking algorithms, and accurately locate objects to make effective collision avoidance decisions.
When choosing a solution, another important factor to consider is the full coverage of the field of view and the high detection rate on all types of surfaces. This is essential for safety systems that need to detect objects around cruising vehicles-low-resolution sensors may be possible Missing measurements will occur.
Finally, the data from the safety CPU will be sent to the navigation CPU and used as redundant data. Redundant sensors improve the performance level, detection rate and robustness of the system. Objects missed by one sensor will be captured by another sensor, thereby significantly improving system performance and ensuring safe operation, which is essential for fully automated unmanned driving involving multiple passengers.
A solution that combines performance and cost advantages while meeting safety requirements
Based on the above requirements and the inherent limitations of sensor technology, many cruise vehicle developers turn to solid-state floodlight LiDAR to eliminate blind spots around cruise vehicles. Compared with mechanical scanning LiDAR, floodlight LiDAR has lower cost and longer mean time between failures (MTBF), can provide highly reliable short-range detection, and can achieve 100% optical density and full field of view coverage. The data provided by floodlight LiDAR can track objects at a measurement rate of up to 100Hz, predict possible collisions based on speed, directionality and location, and eliminate any blind spots that may exist around the vehicle.
When the remote main detection system cannot detect objects due to failure or omission, the LiDAR solution, as a redundant braking and collision avoidance system, will fully consider the maximum deceleration rate (especially the maximum speed of many typical cruising vehicles) Provide the necessary scanning range.
In addition, if an object is detected in the vicinity of the vehicle, an emergency braking procedure will be initiated. In addition, passengers on cruising vehicles often do not wear seat belts and maintain a standing posture. To ensure that the speed remains at a safe level when decelerating, a certain braking distance is required.
For example, if a cruise car traveling at a speed of 40 km/h uses a deceleration rate of 3.5 m/s², it will take 23 meters (including 0.5 seconds of response time) to stop.
In order to cover a sufficient range, many LiDAR solutions require more powerful and expensive optical equipment and laser sources. Through patented technology and software method for signal processing, LeddarTech’s floodlight LiDAR technology has a cost advantage, can provide the required range and performance, in the commercial deployment of cruise vehicle applications (such as emergency braking and blind spot monitoring)
It is extremely attractive.
To ensure coverage of all blind spots, Leddar sensor modules can be used to implement two main types of architecture. If cruise vehicle manufacturers need other levels of resolution, range or coverage, they can choose the corresponding module or architecture according to specific application scenarios. Therefore, if the design team can accept some blind spots, the number of sensors required will be reduced.
1. The first structure provides a short-range safety car shell, which can cover all sides of the cruiser and eliminate blind spots.
To achieve this goal, we adopted eight Leddar M16-LSR LiDAR modules, which have a 100-degree horizontal field of view and a 12-meter pedestrian (low reflectivity) range. These eight modules are directly installed in the body of the cruiser. The height of the ground varies from 20 cm to 75 cm, and small objects in close proximity can be detected.
Figure 1 shows the short-range safety car shell view provided by the first sensor architecture.
2. The second type of architecture uses short-range coverage on the sides and back of the cruiser, and expands the coverage on the front of the car, which can be used for redundant collision avoidance systems. To achieve this goal, we placed five modules with 100-degree FOV on the sides and back of the cruiser, and installed two M16-LSR LiDAR sensors with greater coverage and higher resolution (48-degree FOV, 30 Meters pedestrian range) placed in front of the vehicle body.
Figure 2 shows the short-range car shell view provided by the second sensor architecture, with a longer coverage in front of the car body.
Use LeddarTech’s solid-state LiDAR technology to realize the cruise vehicle safety car shell
The following is a brief introduction of LeddarTech’s 2D floodlight LiDAR solution, which is used by the developers of the autonomous cruise vehicle in the safety car shell and emergency braking.
Leddar M16-LED and M16-LSR modules
The module has two main configuration types according to different lighting methods: classic LED series and new laser series. Both M16 series are very suitable for outdoor operation: they use solid-state design, do not need to install electric machinery, have a wide operating temperature range, adapt to various climates, and are not affected by lighting conditions.
It is the proven main product of Leddar, which is highly favored by customers due to its versatility and reliability. The module’s infrared LED light source can provide wide beam illumination with a maximum range of 100 meters, and has six different field of view configurations. M16-LSR uses a laser source to achieve a longer range, and provides a narrower and more well-defined vertical FOV. Compared with M16-LED, it has a smaller profile.
The main functions and advantages include:
• Excellent angular resolution, using 16 independent segments to track and collect at the same time
• Different beam options to achieve the best field of view (FOV)
• Multi-target tracking and lateral discrimination capabilities
• FOV detection range is up to 165 meters (541 feet), and is related to the target (M16-LSR is relative to the reflective target)
• Fast data acquisition time (up to 100 Hz)
• With various beam options, applicable to various hybrid vehicle configurations
Floodlight LiDAR technology is developing rapidly. Thanks to the in-depth development of the new LiDAR architecture and the continuous improvement of signal processing, newer and more powerful solutions are just around the corner. At the same time, the vast market and a large number of application requirements are also powerful driving factors. In the end, floodlight LiDAR technology will be more cost-effective, and its performance will completely replace mechanical scanners in the next few years, which is exactly what the automotive industry experts expected. Therefore, the flood-lit LiDAR solution, which has been adopted as part of the sensor suite by the automatic cruiser, will play a more important role when the next-generation products are widely used.
Based on LeddarTech’s patented technology, a 3D solid-state floodlight LiDAR with excellent range and resolution will be launched soon, which can meet the needs of advanced automatic cruise detection.