Building a Level 4 (or above) autonomous vehicle makes you strongly dependent on the combination of hardware sensors and the 3D perception software that uses them to detect signs, obstacles, and other vehicles around you.
But when you need to integrate several sensors, which is always the case, managing to position the sensors in a robust manner so the shock and vibration do not destroy your alignment is not obvious. Our L3CAM is a multimodal autonomous vehicle perception system including several different imaging modes combined in a very compact casing. All imaging modes come fused from the factory so early or late data fusion algorithms may be directly applied to 3D perception. Forever forget about recalibrations and checkerboards after bumpy road dataset acquisition. Further, the unit contains an embedded edge computer that can run AI perception directly in the camera. A single Ethernet output cable connects the unit to an external computer.
We have been delivering customized systems for LIDAR autonomous vehicles and building prototypes for different customers in the mobility area, widely understood (ground, sea and air). Autonomous vehicle perception systems in all markets are essentially object detection sensors that pursue comparable goals around vehicle navigation, collision avoidance, object detection and tracking, and advanced surveillance using multisensory approaches.
In automotive, we have built tailored object detection sensors based on the L3CAM technology for data collection under different environments, from sunny roads, adverse weather, or even smoke and fog. Hardware and software developments oriented to 3D LIDAR autonomous vehicles and ADAS autonomous driving, including different ADAS system functionalities, including ADAS LIDAR systems have been implemented in the automotive field.
Similar approaches have been investigated for railways, adapting the system to the requirements of the train for use cases like predictive maintenance, autonomous shunting, or autonomous navigation.
Solutions for the railway industry involving ADAS LIDAR and cameras go beyond autonomous trains. For instance, applications related to prevention or mitigation of collisions in areas with railway or TRAM services mixed with automobiles in urban environments, and monitoring of level crossings, have been developed in the railway sector. In addition, our L3CAM multimodal sensor including high-res LIDAR can also be used for continuous infrastructure monitoring, automated inspection of installations, and preventive inspection of trenches for gauge space validation, including the presence of obstacles, even moving ones.
Multimodal sensors are also of value in several maritime applications. Novel automated cargos are starting to sail, for which the detection of smaller boats or obstacles around them is critical, and can be performed with a sensor fusion suite like the L3CAM. Marine LIDAR imaging sensors, especially if combined with polarimetric imaging, have a special capability to detect small cross-section floating objects (like buoys or surface debris) around the ship, complementing the capabilities of radar sensors and building strong all-weather autonomous vehicle object detection sensors. Similarly, but for different reasons, small recreational or fishery vessels also benefit of 3D perception based on point cloud processing for the detection of small floating obstacles and other vessels as navigation support or for safe autonomous navigation. Another important use case is the detection of person-to-water situations, where sensor fusion LIDAR cameras based on 3D LIDAR can support the detection of persons in the sea even in complete darkness (or in bright sunlight) supported by 3D perception algorithms.
In space, we have been developing high resolution LIDAR mainly for 3D perception for In-Orbit Servicing applications. Robotic and autonomous spacecrafts and satellites are being prepared for the next steps of space exploration and services in a variety of uses, which require high accuracy sensor fusion LIDAR cameras for fail-proof operation, including in particular high resolution 3D LIDAR imaging sensors that are expected to be critical in those purposes. Applications include orbital robotics, all types