Efficient CORDIC Iteration Design of LiDAR Sensors' Point-Cloud Map Reconstruction Technology

Area of Science:

• Electrical Engineering
• Computer Engineering
• Robotics

Background:

• LiDAR sensors are crucial for autonomous vehicles, requiring efficient data processing.
• Existing CORDIC algorithms can be computationally intensive for real-time LiDAR applications.

Purpose of the Study:

• To propose a novel CORDIC iteration circuit design for enhanced LiDAR sensor performance.
• To reduce computational complexity and improve the speed of LiDAR data decoding.

Main Methods:

• Developed a novel CORDIC architecture with pre-selected angles to minimize iterations.
• Implemented and tested the CORDIC circuit for LiDAR packet and 3D frame decoding.

Main Results:

• Achieved trigonometric function calculation in just seven rotations, reducing iterations by over half.
• Experimental results confirmed exact matching with ground truth data.
• The designed chip boasts a small area (1.93 mm × 1.93 mm) and low gate count (129,688).
• Demonstrated rapid decoding times (0.012 ms for packet, 0.912 ms for 3D frame) and high throughput (8.2105 × 10^8 bits/sec).

Conclusions:

• The proposed CORDIC design offers a significant improvement in efficiency for LiDAR sensors.
• This optimization reduces computation time, chip area, and power consumption.
• The efficient CORDIC solution is vital for real-time point-cloud map reconstruction in autonomous driving.