Small Sensor, Big Impact: How Near-Range Lidar Transforms Autonomous Driving

Near-range lidar technology is transforming autonomous driving by enabling precise proximity perception to ensure safe and efficient operation. Often small and discretely embedded, near-range lidar sensors are the hidden hero of autonomous vehicles (AVs), working alongside cameras and radars to create an error-proof safety cocoon.

What are the unique advantages of near-range lidar? In what use cases is it most helpful? What placement strategies should OEMs consider for maximized results?

In this article, Henri Haefner , Senior Director of Product Management of Marketing at Cepton, shares his insights from the forefront of day-to-day near-range lidar implementation:

1. Near-range lidars for autonomous driving differ in performance from those used in ADAS.

For consumer-level deployment, near-range lidars in AVs and Advanced Driver Assistance Systems (ADAS), need to prioritize different performance metrics.

ADAS systems may prioritize resolution over coverage, but as AVs are often required to have comprehensive environmental awareness, near-range lidars for autonomous driving need to map out a broader range of surroundings from ground level, with a large horizontal field of view (FOV) and a vertical FOV no smaller than 75°. This not only helps minimize blind spots in detecting safety-critical objects within the vicinity of the vehicle, but also streamlines integration by reducing the number of sensors needed per vehicle.

Because AVs are likely to be outfitted with multiple sensors all around them, compact sensor designs and low power consumption are also critical factors.

While AVs require robust performance in diverse conditions and environments, ADAS systems can afford to compromise on certain parameters to achieve a balance between performance, cost efficiency and reliability.

However, since AVs are still being developed through smaller, controlled deployments, their requirements for lidar might be different in terms of performance, reliability, embeddability and power consumption.

2. From parking to highway driving, near-range lidar enables a wide range of AV use cases.

Lidar offers high-precision, real-time 3D sensing to significantly enhance the safety and efficiency of AV operation. Near-range lidar focuses on enabling high-fidelity perception of a vehicle’s immediate surroundings. Some of its key use cases include:

• Valet parking and obstacle detection: Near-range lidar sensors offer superior capabilities of detecting small and dark obstacles in scenarios where human or camera vision is limited. This makes them very useful in valet parking and maneuvers through tight spaces.
• Highway driving: Near-range lidar sensors, when embedded on the sides of vehicles, can significantly improve critical highway maneuvers, such as lane change, to ensure safe merging into traffic. This is due to their small form factor, low cost and high resolution.
• Blind spot minimization: Near-range lidar plays a pivotal role in reducing perception blind spots in urban driving, especially in right turns or in start-stop scenarios. The safety features that it enables are invaluable in large commercial vehicles, where mirrors and cameras fail to offer visibility that is high enough to alert the vehicle of all potential hazards.

3. Near-range lidar has distinct advantages over other sensor technologies across many critical scenarios.

Near-range lidar has much higher special resolution than radar and sonar, enabling the accurate detection of small obstacles. This results in significantly lower false positives and false negatives. Combined with its wide FOV coverage, lidar enhances the sensor suite when working alongside radar.

Compared to camera and computer vision, near-range lidar has unique strengths thanks to its ability to perform in challenging lighting conditions and provide intrinsically 3D data. This makes lidar extremely useful in detecting obstacles at night or in strong sunlight, while providing accurate information about their size, distance and velocity.

The integration of lidar, radar and camera is crucial for enhancing the safety and functionality of AVs across various conditions and driving scenarios. Together, they deliver effective, high-precision 3D perception, with an optimized combination of computational resource, power efficiency, system footprint and cost.

4. Near-range lidar placement in AVs varies by vehicle size and type.

For commercial vehicles like trucks and vans, near-range lidars often have a higher mounting position. This is intended for increasing the sensor FOV and minimizing potential damage from road debris. An elevated location optimizes visibility for urban and highway driving scenarios.

Passenger vehicles might benefit from an integration plan of near-range lidar sensors that balances functionality with modern vehicle aesthetics. Locations such as side mirrors, headlamps or front grille are popular as they allow the lidar sensors to be discretely hidden while ensuring optimal aerodynamics. These locations also offer good sensor coverage for common use cases such as parking and obstacle detection.

Near-range lidar placement strategies vary from OEM to OEM, depending on each vehicle’s specific design requirements and use cases. Automakers need to work closely with lidar suppliers to come up with a highly customized plan to achieve the right balance between performance, aesthetics and functionality. Naturally, it would cost a lot of money and time to physically install and test lidars in different locations of the vehicle. Therefore, true-to-life lidar simulation provides a potent solution that virtually visualizes and validates integration options across different lidar models, placement locations, sensor orientations and vehicle types.

5. Rigorous hardware testing is required for deploying near-range lidar and maximizing its benefits.

To ensure the effective and safe implementation of near-range lidar in AVs, rigorous testing must be implemented at multiple levels:

• Performance verification that evaluates range performance, resolution capabilities, detection and measurement accuracy and diagnostic features, to identify and address potential issues.
• Component-level testing assesses reliability and durability of the lidar sensors, regardless of system requirements. Common testing scenarios, such as random vibration, salt spray, high humidity and temperature, and temperature shock, ensure that the components are robust enough for daily operation and maintain consistent performance over time.
• System-level testing that aligns with the specific Operational Design Domain (ODD) of the AV. And verifies performance across various environmental conditions and driving scenarios that are specific to the intended use. Commonly tested scenarios include lane changes, emergency braking, highway merging and small obstacle detection, either on proving grounds or in simulated environments.

6. Software is also essential to lidar’s safe deployment in AVs.

Software support needs to be properly implemented to ensure or enable functional safety, cybersecurity and perception.

• Low level: fundamental functions such as the generation of calibrated point clouds, range measurement and potential obstruction detection within the lidar FOV.
• Mid-level: more advanced capabilities, including ground detection, ego-motion (estimating the vehicle’s own movement), real-time calibration adjustments and occupancy grid generation for spatial awareness.
• High-level: features that make the lidar data more intelligent, such as tracking and classifying objects, generating detailed object lists and semantic labeling.

To learn more about near-range lidar for AV, visit our blog post “The Hidden Hero: Near-Range Lidar in Autonomous Vehicles”.