Point Cloud Geometry Encoding With Sensing Coverage Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing point cloud compression technologies, such as G-PCC, struggle with efficient encoding and decoding of point cloud geometry data from sensors with flexible sensing paths, particularly in low-latency applications like autonomous driving, where mechanical spinning sensors are prone to failure and limited in view, and there is a lack of reliability indicators for data transmission errors.
Innovation Solution
The method involves encoding and decoding point cloud geometry data using a coarse representation in a two-dimensional space defined by sensor and sample indices, incorporating sensing coverage data to indicate missing data and improve reliability, allowing for efficient compression and decoding even with unreliable transmissions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If mechanical spinning sensors are used for point cloud sensing, then the sensor can cover a wide view, but the sensor is prone to failure and has limited reliability
Solution Approach 1:
The patent segments the sensing task across multiple sensors (e.g., multiple fixed sensors or a combination of fixed and mobile sensors) rather than relying on a single mechanical spinning sensor. This distribution of sensing functions improves system reliability while maintaining comprehensive coverage area.
Solution Approach 2:
The patent replaces mechanical spinning sensors with non-mechanical sensing approaches, such as electronic beam steering or multiple fixed sensors, eliminating the mechanical failure points while maintaining wide sensing coverage through coordinated operation.
2Productivity
If traditional point cloud compression methods are used, then encoding and decoding can be performed, but transmission errors cannot be detected and reliability indicators are lacking
Solution Approach 1:
The patent incorporates feedback mechanisms where the encoder provides sensing coverage information to the decoder, enabling the decoder to detect transmission errors by comparing expected versus received data patterns. This feedback loop maintains encoding efficiency while adding reliability detection capabilities.
Solution Approach 2:
The patent performs preliminary encoding of sensing coverage metadata alongside the point cloud data, preparing reliability indicators in advance before transmission. This allows the decoder to immediately detect errors without additional processing delays, maintaining productivity while enhancing reliability.
3Measurement precision
If complete point cloud data is transmitted to ensure accuracy, then decoding accuracy is improved, but energy consumption increases
Solution Approach 1:
The patent extracts and transmits only the essential sensing coverage metadata and reliability indicators separately from the complete point cloud data. This allows the receiver to detect errors and assess data quality without transmitting unnecessary data, reducing energy consumption while maintaining decoding accuracy for the transmitted portion.
Solution Approach 2:
The patent transmits a partial representation of the point cloud data along with sensing coverage information, which is sufficient for error detection and quality assessment. This partial transmission approach reduces energy consumption while the sensing coverage data enables accurate reconstruction and error detection.
Data Source
AI summary
A method of encoding, into a bitstream, point cloud geometry data sensed by at least one sensor associated with a sensor index is provided, the point cloud geometry data being represented by ordered coarse points occupying some discrete positions of a set of discrete positions of a two-dimensional space, each coarse point being located within the two-dimensional space by an order index defined from a sensor index associated with a sensor that is able to sense a point of the point cloud represented by the coarse point and a sample index associated with a sensing time instant at which the point of the point cloud is sensed. The method includes: obtaining sensing coverage data (SCD) representative of at least one range of order indexes associated with sensed data; and encoding the sensing coverage data (SCD) into the bitstream.