Velodyne Lidar has released the latest version of our Vella Development Kit (VDK) – called 2022.1 – that allows customers to plug in our lidar with an off-the-shelf library of perception capabilities. The VDK advances development for ever-evolving autonomous applications and accelerates time to market.
The VDK makes it easy for developers to pair our breakthrough Vella perception software with our world-class lidar sensors to speed innovation cycles for bringing full-stack solutions to market. For example, VDK users can build an autonomous mobile robot (AMR) solution with Vella software that utilizes data from our Velarray M1600 sensor to detect people and any 3D obstacles. These obstacles include overhanging cables, small objects such as dunnage on the floor and changes in the floor like ramps, staircases and small holes. For a last mile delivery application, the sensor data provides object and hazard detection to protect roadway users including pedestrians, bicyclists and pets.
The VDK advances solution development for bringing groundbreaking lidar perception capabilities to mobile delivery devices, industrial robotics, drones, autonomous vehicles, advanced driver assistance systems (ADAS) and more. The new release demonstrates our ongoing commitment to update the VDK with the latest hardware support and processing features that Vella has to offer.
Existing customers have used our VDK for short-range pedestrian detection, cross-reference and fusion with camera-based perception for ADAS applications, collision avoidance for industrial applications and more.
Included among the new VDK features are:
The VDK has a new version of its machine learning models, which are more robust and yield higher performance than its predecessor due to more learning of corner cases. The models can detect objects in multiple roadway scenarios including urban, suburban and highways. They can also provide point-wise classification for different categories including drivable surface, sidewalks, terrains, movable and static obstacles. The improved reliability of machine learning perception models allows customers to achieve high performance metrics for their applications. Previously our customers had to spend significant resources and time to deploy lidar in their solutions. With VDK’s advanced lidar perception capabilities, the customers’ time to market is accelerated and they can focus on their domain or application.
Lidar odometry has been added to VDK in this release. This capability uses lidar to estimate automated vehicle (or ego) motion which is crucial for path planning. The lidar odometry also enables the users to accumulate lidar frames to gain higher resolution and density for static objects. Lidar odometry paves the way for rapid and easy integration of VDK across various industrial applications. It is agnostic to various environments and thus can work in a range of applications and industries. In particular, this VDK technology is a must-have capability for mobile platforms providing autonomous solutions, allowing customers to easily add lidar odometry to their applications.
The VDK includes support for ARM-based hardware, specifically the NVIDIA Jetson Xavier AGX module, which provides unparalleled computing power in a compact, energy-efficient module. The GPU-accelerated module enables specific VDK algorithms to processes lidar frames in real time to detect, classify and extract objects.
ARM-based support allows developers of industrial automated guided vehicles (AGVs) and automated mobile robots (AMRs) to integrate VDK into their systems due to the smaller form factor as well as low power requirements. It’s challenging to maintain high accuracy and fidelity of different capabilities while optimizing the run-time performance on lower powered devices. With continuous optimizations of our algorithms, the VDK in the coming releases will be supported on additional low-powered compute platforms.
The VDK’s 2022.1 updates significantly enhance the software toolkit by updating machine learning models and adding lidar odometry which helps to remove any dependency on integration with odometry sensors on the platform. The VDK also supports both ARM64 and AMD64 platforms.
The VDK provides a great way for developers to experience Velodyne’s software capabilities and advance autonomous system innovation by developing full-stack solutions using our software and sensors.
To learn more about Velodyne’s VDK Platform, reach out to [email protected]