Published:
A new method for tuning noise in UAV AHRS estimation uses Bayesian optimization for fast, efficient improvements. By adjusting scaling factors of noise matrices, it efficiently optimizes for high-dimensional systems. Performance tests show significant improvement with this method, which is widely applicable to AHRS applications.
Published:
This project focuses on creating a reliable Attitude and Heading Reference System (AHRS) algorithm using real-time data from the ICM-20948 sensor. Specifically designed firmware code was developed for deployment on the TMS320F28377S Texas Instrument processor, enabling the achievement of a high accuracy AHRS estimation at an impressive output rate of 110 Hz. The demonstration in the provided video showcases this accuracy. Moreover, a pivotal feature of the project involves implementing data export in the MAVLINK protocol, facilitating seamless interfacing for UAV and robotics applications, thereby enhancing accessibility and usability within these domains.
Published:
Drone navigates indoors without GPS: This project equipped a drone with stereo cameras, lidar, and ultrasonic sensors to navigate autonomously in GPS-denied environments. Cameras + lidar + IMU helped track the drone’s location, while ultrasonic sensors scanned for obstacles. The project even upgraded to fly through windows, using YOLO3 for object detection and obstacle avoidance.
Published:
The project focuses on implementing drone obstacle avoidance using lidar sensor data. During autonomous navigation, the drone identifies obstacles along its path, promptly adjusting its course to navigate around them and ensure an unobstructed trajectory towards mission completion.