Point Cloud Compression for Low-Latency Cooperative Perception
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous air vehicles (AAVs) face limitations in visibility due to occlusions, sensing range, and extreme weather, and existing data sharing methods result in excessive data traffic and loss of important information, necessitating efficient point cloud compression and intelligent cooperative perception systems.
Innovation Solution
Implementing a method and system that includes information-centric networking (ICN) for flexible communication, deep reinforcement learning for adaptive data selection, and Recurrent Neural Network (RNN)-based point cloud compression to optimize data transmission and fusion, ensuring accurate and fast object detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If AAVs share complete point cloud data from all sensors, then object detection precision is improved, but data traffic volume becomes excessive for 5G network capacity
Solution Approach 1:
The patent extracts only the essential and important information from complete point cloud data for transmission. This includes selectively transmitting point clouds based on their importance to object detection, rather than transmitting all sensor data. The system identifies and transmits only the critical portions of point cloud data that contribute to accurate object detection, thereby reducing data traffic volume while maintaining detection precision.
Solution Approach 2:
The patent applies different quality levels to different portions of point cloud data based on their importance. Critical regions containing objects of interest are transmitted with high quality and detail, while less important background regions are transmitted with lower quality or omitted entirely. This local quality differentiation ensures that network bandwidth is allocated efficiently to the most important data while reducing overall traffic volume.
2Quantity of substance
If AAVs transmit only object types and locations, then data traffic volume is reduced, but important sensory information is lost
Solution Approach 1:
The patent transmits detailed point cloud data locally for regions containing objects of interest, while using simplified representations for other areas. This ensures that important sensory information about objects is preserved in the transmitted data, while reducing overall information loss by not requiring complete high-fidelity data for all regions.
Solution Approach 2:
The system performs preliminary analysis of point cloud data to identify regions containing objects of interest before transmission. This preliminary action allows the system to pre-select and prioritize which data portions require detailed transmission versus which can be summarized, thereby preserving important information while minimizing data volume.
3Measurement precision
If complete point cloud data is transmitted in real-time, then cooperative perception accuracy is improved, but network latency increases
Solution Approach 1:
The patent extracts and transmits only the critical point cloud data necessary for cooperative perception, rather than transmitting complete datasets. By identifying and transmitting only the essential portions of point cloud data that contribute to perception accuracy, the system reduces transmission time and network latency while maintaining cooperative perception performance.
Solution Approach 2:
The system transmits a partial set of point cloud data that is sufficient for achieving the required cooperative perception accuracy, rather than transmitting excessive complete data. This partial transmission approach reduces network latency while maintaining the necessary perception accuracy for safe autonomous operation.
4Productivity
If intelligent data selection is implemented, then data transmission efficiency is improved, but system complexity increases
Solution Approach 1:
The patent implements preliminary analysis and classification of point cloud data to identify important regions and objects before transmission. This preliminary action enables intelligent data selection based on pre-established criteria and algorithms, improving transmission efficiency while managing system complexity through structured preprocessing rather than complex real-time decision-making.
Data Source
AI summary
A method for point cloud compression of an intelligent cooperative perception (iCOOPER) for autonomous air vehicles (AAVs) includes: receiving a sequence of consecutive point clouds; identifying a key point cloud (K-frame) from the sequence of consecutive point clouds; transforming each of the other consecutive point clouds (P-frames) to have the same coordinate system as the K-frame; converting each of the K-frame and P-frames into a corresponding range image; spatially encoding the range image of the K-frame by fitting planes; and temporally encoding each of the range images of the P-frames using the fitting planes.