RS-LiDAR-Algorithms

RS-LiDAR-Algorithms is a SDK that RoboSense specially developed for Autonomous Driving Applications. Packed in the SDK are algorithm modules including localization, road curbs/driving area detection, lane markings detection, obstacles detection/classification, and moving objects tracking, etc. The purpose is to facilitate client’s secondary development and speed up their autonomous driving projects.

High Precision Localization

High precision localization is the premise for autonomous driving environment perception. RS-LiDAR-Algorithms comprises high precision real-time localization modules of industry leading localization precision(≤20cm), to surpass the requirements of autonomous driving.

X coordinate:277

Y coordinate:360

Road Curbs /Driving Area Detection

Driving area detection lays the groundwork for autonomous path planning. RS-LiDAR-Algorithms offers modules of road curbs detection and driving area detection to provide pathfinding function.

Lane Markings Detection

Like road curbs, lane markings is another important information that is indispensable in path planning. RS-LiDAR-Algorithms contains lane markings detection module, which can precisely extract road sign information including lane markings, road signs, zebra crossing etc. by analysing the different reflective intensity information contained in the returned laser signals.

Obstacles Detection

Obstacles detection is the basic needs for safe drive and is a way of interaction between autonomous cars on the road. RS-LiDAR-Algorithms provides obstacles detection module which can detect and output in real time the location, distance, position, size and shape information of multiple obstacles to help the cars “understand” the surroundings and make decisions.

x:299 y:99

78m

RoboSense

x:210 y:33

100m

x:122 y:101

92m

Moving Objects Tracking

Autonomous cars need to pay extra attention to moving objects. RS-LiDAR-Algorithms has included moving objects tracking module which can estimate and output in real time the moving information of multiple objects including the speed, size and direction. The speed information can be further analysed to speculate the acceleration and angular speed of objects which will indicate their trajectory and behavior intentions.

36km/h

truck

47km/h

RoboSense

1km/h

people

57km/h

car

2.1km/h

people

Obstacle Classification/Recognition

By classifying objects, the autonomous driving system can better estimate the behavior intentions of nearby objects and make more precision path planning and control decisions. RS-LiDAR-Algorithms has obstacles classification and recognition modules which can catalogue obstacles into pedestrians, bicycles, automobiles, trucks and more. To promote objects recognition precision, RoboSense has constructed an industry leading LiDAR scene data base.

truck

RoboSense

people

car

bike