3D Motion Grid Occupancy Mapping for LIDAR Occlusion Handling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicle systems face challenges in accurately segmenting objects from LIDAR point clouds due to occlusions and blind spots, leading to phantom movements and incorrect velocity estimates, which can result in inappropriate responses to actual objects or failure to respond to existing objects.
Innovation Solution
A 3D motion grid system processes LIDAR data to create a 3D point cloud, which is discretized into top-down bird's eye view grids, using ray tracing to characterize grid cells as 'freespace,' 'occupied,' 'ground,' or 'overhang,' and tracks motion by maintaining a temporal rolling window, enabling classification of objects and their motion, and utilizing a GPU for real-time processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LIDAR sensors are used to detect objects in autonomous vehicles, then object detection capability is improved, but occlusions and blind spots cause phantom movements and incorrect velocity estimates
Solution Approach 1:
The patent combines data from multiple LIDAR sensors positioned at different locations on the vehicle to create a unified point cloud representation. By merging the detection fields of multiple sensors, the system eliminates blind spots and reduces phantom movements caused by individual sensor limitations, thereby improving both measurement precision and reliability simultaneously
Solution Approach 2:
The patent introduces a motion grid system as an intermediary layer between raw LIDAR data and object classification. This motion grid acts as a mediator that processes point cloud data to identify true motion patterns, filtering out phantom movements and providing more reliable velocity estimates for downstream processing
2Area of stationary object
If multiple LIDAR sensors are deployed to reduce blind spots, then coverage area is improved, but data processing complexity increases
Solution Approach 1:
The patent segments the detection space into a motion grid structure, dividing the complex point cloud data from multiple LIDAR sensors into manageable grid cells. Each grid cell can be processed independently, reducing the overall computational complexity while maintaining comprehensive coverage area
Solution Approach 2:
The patent transforms the three-dimensional point cloud data into a two-dimensional motion grid representation from a top-down perspective. This dimensional transformation simplifies the processing of multi-sensor data by projecting spatial information onto a 2D plane, making it more computationally efficient while preserving detection coverage
3Speed
If real-time processing is implemented for autonomous vehicle perception, then response time is improved, but computational load increases
Solution Approach 1:
The patent extracts and processes only the most relevant features from the LIDAR point cloud data using the motion grid system. By taking out and focusing on critical motion patterns rather than processing all raw data points, the system achieves real-time processing speeds while reducing computational energy consumption
Solution Approach 2:
The patent implements partial processing by maintaining a temporal rolling window that processes only recent and relevant frames rather than the entire data history. This partial action approach enables real-time response while keeping computational energy consumption within acceptable limits
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The system effectively identifies and classifies objects and their motion, reducing phantom movements and incorrect associations, enabling accurate detection and response to actual objects, and optimizing processing for real-time operation on a GPU.
Implementation Method 1
LIDAR (light detection and ranging) sensor that detects and ranges electromagnetic radiation, such as laser beams
Data Source
AI summary
A method is described and includes receiving a frame of point cloud data from at least one onboard light detection and ranging (LIDAR) sensor of a vehicle; discarding points of the received frame of point cloud data that are outside a defined geographic area around the vehicle; and, subsequent to the discarding, performing ray tracing in connection with the remaining points of the received frame of point cloud data. The method may further include characterizing an occupancy condition of each of a plurality of cells of a three-dimensional (3D) grid corresponding to the defined geographic area based on the ray tracing of the received frame of point cloud data, wherein the 3D grid corresponds to the received frame of point cloud data.