Back to Projects
High Speed Drone Stereo Pipeline
Fall 2025 - PresentROS2
Python
Docker
cuVSLAM
rectification
Overview
The goal of this project was to create a high speed stereo depth estimation pipeline for a high speed drone with inputs being stereo images and outputs being pose and depth maps. This project was done on a Jetson Orin Nano realtime with latency requirements.
The Challenge
Some notable challenges were: latency, benchmarking, integration, etc.
Key Features
- Runs on Jetson Orin Nano at 15 Hz
Outcome
Learned a lot about integrating software on real time systems