As autonomous vehicles (AVs) evolve, the integration of lidar alongside radar and cameras becomes pivotal in achieving robust and foolproof autonomy. Near-range lidar technology is revolutionizing automated driving capabilities, offering precise environmental sensing that is crucial for safe and efficient operation.
In this blog, we will explore the distinct advantages of near-range lidar, its common use cases, optimal placement strategies and essential considerations for effective implementation in AV systems.
1. Do lidars for AV and ADAS entail different performance and features?
Near-range lidar systems designed for consumer-level AVs and Advanced Driver Assistance Systems (ADAS) indeed prioritize different performance metrics and features.
One of the main differences lies in Field of View (FOV). AVs typically require comprehensive coverage capabilities, which entails a large horizontal FOV and at least a 75° vertical FOV to effectively monitor surroundings from ground level to the horizon. The extensive FOV not only enhances safety by detecting obstacles and pedestrians but also reduces the number of sensors needed per vehicle, optimizing integration and cost efficiency. Moreover, due to the need for 360° coverage around the vehicle, AVs benefit from compact designs and low power consumption to accommodate multiple sensors.
ADAS systems may prioritize high-resolution sensing over wide coverage, focusing on precise detection and classification of nearby objects, especially within shorter ranges. While AVs demand robust performance across various conditions and environments, ADAS systems may tolerate compromises in certain parameters to achieve cost-effectiveness without compromising safety and functionality.
However, since AVs are still currently implemented in smaller volumes and in controlled deployment with access by the fleet owner, they may not require the same lidar performance, reliability, embeddability and power efficiency as when they are deployed in mass-market consumer vehicles. That’s why we can still see most AV fleets equipped with bulkier or spinning lidar units on the road.
2. What are common use cases for near-range lidar in AV? Lidar technology plays a pivotal role in advancing autonomous driving capabilities, offering precise and real-time environmental sensing that is essential for safe and efficient operation. From parking to highway driving, near-range lidar enables a wide spectrum of use cases, including:
Valet parking and obstacle detection: Near-range lidar is crucial for detecting small and dark obstacles in the vehicle's immediate surrounding, ensuring safe navigation during maneuvers such as valet parking or navigating through tight spaces where visibility may be limited.
Highway maneuvers: Near-range lidars can be embedded on the sides of vehicles due to their smaller size and lower cost, supporting lane change maneuvers on highways by detecting nearby vehicles and ensuring that adjacent lanes are clear. With high resolution, they enhance the AV's ability to safely merge into traffic and maintain situational awareness.
Blind spot detection: Near-range lidar is invaluable for detecting objects in perception blind spots, particularly during right turns or in start-stop scenarios common in urban environments. This capability improves safety for large commercial vehicles by alerting drivers to potential hazards that may not be visible through conventional mirrors or cameras.
3. What is near-range lidar’s advantage over other sensor technologies?
Near-range lidar offers distinct advantages over other sensor technologies such as cameras, radar, and sonar in several critical use cases for AVs.
Unlike radar and sonar, which may suffer from lower resolution, near-range lidar detects small obstacles with significantly improved precision, reducing false positives and false negatives. Along with its wide FOV coverage, lidar performs significantly better than radar alone.
Lidar is able to operate effectively in poor lighting conditions and to discern dark or low-contrast objects from the ground, which camera struggles with due to its dependence on external illumination. Lidar is also a 3D sensor, while camera provides 2D pictures that can only indirectly (and often inaccurately) provide depth information with the help of computer vision.
These advantages underscore near-range lidar's pivotal role in enhancing AV safety and functionality in diverse environmental conditions and operational scenarios. Lidar, radar and camera complement a powerful sensor suite to deliver effective, high-precision 3D perception for AVs to navigate safely and autonomously, while enabling an optimized combination of computational resource, power consumption, sensory system footprint and cost.
4. Why is Nova an ideal solution for near-range sensing in AV?
Nova stands out as an ideal solution for near-range lidar in AVs due to its compact size, low power consumption and high FOV coverage both horizontally and vertically. Its small form factor allows for seamless integration into multiple locations around the vehicle, without compromising aesthetics or aerodynamics. Its minimized power consumption optimizes energy efficiency, crucial for the continuous operation of AV systems. Its 120° x 90° FOV is horizontally wide enough and vertical tall enough to ensure comprehensive coverage around the vehicle, forming a virtual safety cocoon in the vehicle’s close proximity.
These features are enabled by Cepton’s patented lidar architecture and proprietary System on a Chip (SoC), which maximizes the lidar’s efficiency and data processing capabilities without increasing sensor footprint and power consumption. This makes Nova a versatile and effective choice for enhancing the perception capabilities of AVs, supporting reliable and responsive autonomous driving experiences.
5. What would be common placement options for near-range lidars in AV?
Common placement options for near-range lidars in AVs vary depending on the vehicle's size and type.
For commercial vehicles such as vans or trucks, lidars are often mounted at higher positions to minimize potential damage from road debris and to provide an optimal FOV for detecting obstacles and pedestrians. This elevated placement ensures good visibility and perspective, crucial for safety and maneuverability in urban and highway settings.
In contrast, sedans typically integrate near-range lidars in locations that balance functionality with aesthetic considerations. These may include positions near or within the side mirrors, integrated into the headlamp assemblies, or discreetly within the front grille. These placements maintain the vehicle's aerodynamics while ensuring effective sensor coverage for near-range sensing tasks such as parking assistance and obstacle detection.
Ultimately, the choice of lidar placement is tailored to each vehicle type's specific design requirements and intended use cases. OEMs need to work with lidar manufacturers on integration options extensively to optimize performance and aesthetics while meeting the demands of autonomous driving functionality – this sometimes would result in significant physical implementation costs. This is why we developed our own lidar simulation platform, StudioViz, which allows our partners and customers to explore the vast possibilities of integration options across different lidar models, placement locations, sensor orientations and vehicle types. Here’s an example:
6. How to effectively and safely implement lidar on AV to maximize its benefits?
To effectively and safely implement lidar on AVs and maximize its benefits, rigorous testing needs to be implemented on multiple levels.
Performance Verification: Essential tests include evaluating range performance, resolution capabilities, accuracy of detection and measurement and diagnostic features that aid in identifying and addressing potential issues.
Component Level: Regardless of system requirements, lidar components must undergo rigorous reliability and durability testing. Common testing scenarios include random vibration, salt spray, high humidity and temperature, temperature shock and so on. This ensures that they can withstand the rigors of daily operation and maintain consistent performance over time.
System Level: Testing at the system level should align with the specific Operational Design Domain (ODD) of the AV. This includes verifying performance across various environmental conditions and driving scenarios relevant to the intended use. Tested scenarios such as lane change, emergency break, highway merge and small obstacle detection are often conducted on test grounds and in simulation environments.
Additionally, software is an essential enabler of lidar’s safe deployment in AVs, spanning functional safety, cybersecurity and perception.
Low-Level Software Features: These include fundamental functions such as generating calibrated point clouds, estimating range to detected objects, and detecting potential obstructions or blockages in the sensor's FOV.
Mid-Level Software Features: This level encompasses more advanced capabilities such as ground detection, estimation of the AV's own movement (ego-motion), real-time calibration adjustments, and generating occupancy grids for spatial awareness.
High-Level Software Features: At this level, lidar data is processed to generate detailed object lists, track and classify detected objects (like vehicles, pedestrians or cyclists), and apply semantic labeling to understand the environment comprehensively.
By adhering to thorough testing protocols and leveraging comprehensive software support across all levels, AV developers can ensure that lidar systems perform optimally, enhancing the vehicle's perception capabilities and overall safety in autonomous operation. These practices not only validate functionality but also contribute to building trust in lidar technology as a cornerstone of AV development.
Near-range lidar is becoming increasingly adopted by OEMs to advance autonomous driving capabilities. Its ability to provide high-resolution, 3D perception in various environmental conditions sets it apart from other sensor technologies. By strategically placing lidar sensors in use case and vehicle type-specific positions and rigorously testing their performance, AV developers can harness its full potential to enhance safety, optimize vehicle operations, and pave the way for widespread adoption of autonomous vehicles.
