Vision AI-Kit Controls LeKiwi Mobile Robot
A hands-on demo details using the Vision AI-KIT 6490 (QCS6490) with ROS2 Jazzy and a Host PC (Ubuntu 24.04) to control a LeKiwi mobile base via a computer-vision hand controller. It uses MediaPipe v0.10 full models accelerated by Qualcomm AI Runtime (QAIRT), boosting inference from ~10 fps to over 80 fps and mapping hand positions to ROS Twist commands. The distributed setup splits perception on-device and simulation on the host.
Key Points
- 1Accelerates MediaPipe hand-detection to ~80 fps on QCS6490 using Qualcomm AI Runtime (QAIRT)
- 2Deploys distributed ROS2 Jazzy system across Vision AI-KIT and Ubuntu 24.04 host for simulation
- 3Enables practitioners to map hand dials to ROS Twist messages for teleoperation of LeKiwi
Scoring Rationale
High actionability and credible engineering detail for edge-robotics teleoperation; novelty is limited to integration specifics rather than new algorithms.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
