Solid State Lidar 4D Object Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for dynamic object classification in 3D environments, such as autonomous vehicles, require multiple image frames to identify regions of interest (ROI) and track fast-moving objects, which can be inefficient and prone to errors due to the complexity of mechanical motion and analog control systems in optical phase arrays.
Innovation Solution
A photonic integrated circuit-based Lidar system with a 2D beam scanner and coherent optical receiver, using a single image frame to identify ROIs and track objects by generating 4D point cloud data with radial velocity information, enabling efficient object classification and tracking without mechanical motion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple image frames are used to identify ROIs and track fast-moving objects, then object classification accuracy can be improved, but processing time increases and the system becomes less efficient
Solution Approach 1:
The system performs preliminary velocity measurement on all detected objects using single-frame 4D Lidar data. Objects with velocity above a threshold are pre-identified as fast-moving and marked for priority tracking. This preliminary classification enables the system to focus computational resources on only those objects requiring multi-frame tracking, reducing overall processing time while maintaining accuracy for both static and dynamic objects.
Solution Approach 2:
The system dynamically adjusts its processing strategy based on object velocity characteristics. For fast-moving objects identified in the single frame, it activates multi-frame tracking mode to maintain accuracy. For static or slow-moving objects, it uses single-frame processing. This dynamic adaptation allows the system to optimize the trade-off between processing speed and classification accuracy for different object types within the same scene.
2Area of stationary object
If mechanical motion systems are used for scanning, then field of view coverage can be achieved, but system complexity and reliability deteriorate due to moving parts
Solution Approach 1:
The patent replaces mechanical scanning systems with a static optical array consisting of multiple fixed optical elements (lenses, mirrors, detectors) arranged in a specific geometric configuration. Each optical element captures light from a specific angular direction, and the combined data from all elements provides comprehensive field of view coverage without any moving parts. This substitution eliminates mechanical complexity while maintaining full scene coverage capability.
Solution Approach 2:
The system transitions from 2D spatial scanning (requiring mechanical motion) to 4D data capture (adding velocity and time dimensions) using a static array. By measuring both the intensity and phase of light across multiple fixed sensors simultaneously, the system achieves comprehensive spatial coverage and velocity measurement without mechanical scanning, resolving the contradiction between FOV coverage and system simplicity.
3Ease of operation
If analog control systems are used in optical phase arrays, then beam steering can be achieved, but reliability decreases due to suppression of side lobes and large far field beam size
Solution Approach 1:
The patent replaces analog phase control systems with a digital sensor array approach. Instead of using analog modulators to steer beams and suppress side lobes, the system uses a fixed array of optical sensors with known geometric positions. Each sensor detects light from a specific direction, and digital signal processing reconstructs the complete scene including velocity information. This substitution eliminates the reliability issues of analog control while maintaining beam steering functionality through the geometric arrangement of sensors.
Solution Approach 2:
The system creates multiple copies of the optical detection function across fixed sensor elements rather than using a single controllable beam. Each sensor element acts as an independent copy of the detection function, capturing light from its specific angular direction. This parallel copying approach achieves comprehensive coverage and velocity measurement without requiring analog beam steering control, improving reliability by eliminating side lobe suppression issues and far field beam size problems.
4Productivity
If a single image frame is used for object identification, then processing speed improves, but measurement precision for fast-moving objects deteriorates
Solution Approach 1:
The system performs preliminary velocity estimation for all objects using single-frame 4D Lidar data. Objects identified as fast-moving through this preliminary analysis are then subjected to enhanced multi-frame tracking. This two-stage approach allows the system to maintain high processing speed for the majority of objects while applying precision tracking only where necessary, resolving the contradiction between overall processing speed and tracking accuracy for fast-moving objects.
Solution Approach 2:
The system dynamically selects the processing mode (single-frame or multi-frame) based on object velocity characteristics. For fast-moving objects, it activates multi-frame tracking to maintain precision. For other objects, it uses single-frame processing to maintain high throughput. This dynamic processing strategy enables the system to optimize the balance between processing speed and measurement precision adaptively for different objects in the scene.
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
This approach allows for accurate and efficient object classification and tracking using a single image frame, reducing processing time and improving accuracy by leveraging coherent Lidar technology and photonic integrated circuits to generate 4D data without mechanical scanning.
Implementation Method 1
scanning through optical switching across an array of pixels that direct and receive light from a particular angle to an optical chip holding the array of pixels
Implementation Method 2
coherent, frequency modulated continuous wave lasers and corresponding detection to obtain position and velocity information
Implementation Method 3
coherent, frequency modulated continuous wave lasers and corresponding detection to obtain position and velocity information
Data Source
AI summary
Devices are provided to perform imaging using laser light based on scanning without any mechanically moving parts to obtain a scan over the field of view. An optical chip comprises a row of selectable emitting elements comprising: a row feed optical waveguide, a plurality of selectable, electrically actuated solid state optical switches, a pixel optical waveguide associated with each optical switch configured to receive the switched optical signal, and a solid state first vertical coupler associated with the pixel waveguide configured to direct the optical signal out of the plane of the optical chip. The optical chip can be connected with an electrical circuit board to control operation of the optical chip. A lens can be positioned to direct the light from a selected pixel along a specific direction such that a scan over an array of pixels covers a desired portion of the field of view.