Detect indoor objects in real-time from RGB camera input using OmDet-Turbo.
npx skills add https://github.com/bensonlee5/openral --skill omdet-turbo-indoor
Please help me install this Agent Skill. Skill: omdet-turbo-indoor Source: https://github.com/bensonlee5/openral/tree/main/rskills/omdet-turbo-indoor Command: npx skills add https://github.com/bensonlee5/openral --skill omdet-turbo-indoor
This Skill automates object detection in real-time for indoor environments, providing a comprehensive view of the scene without the need for manual prompting.
Deploy the object detection skill on the camera tee in simulation:
openral deploy sim \
--object-detector-manifest rskills/omdet-turbo-indoor/rskill.yaml
High-intent search queries and answers about installing and using this skill.
You can deploy the object detection skill in a simulation environment using the OpenRAL deploy command. It utilizes the OmDet-Turbo model on an RGB camera feed to detect and classify objects within the OpenRAL framework.
Yes, OmDet-Turbo supports an open vocabulary that extends beyond standard COCO categories. It specifically identifies a curated list of indoor items including kitchenware, appliances, and furniture for robot perception.
Yes, you can use OpenRAL to run object detection in-process. The OmDet-Turbo detector executes within the OpenRAL runtime to provide efficient scene understanding and perception for robots navigating indoor spaces.
Real-time detection for robot navigation identifies a fixed vocabulary of indoor objects such as kitchenware, food, appliances, and furniture. This aids robots in tasks like inventory management and object manipulation.
The limitation of fixed vocabulary detection is that it only recognizes a curated list of specific indoor items like kitchenware and furniture. It may not classify objects outside this predefined list without open vocabulary configuration.
bensonlee5Monitor robotic simulation task progress and success from RGB video frames.
ViewWayParse images, PDFs, and scans into labeled regions with bounding boxes.
bensonlee5Automate Franka Panda robot tasks with NVIDIA GR00T N1.7 VLA policy.
GarethManningDesign three-phase outdoor learning sequences linking preparation, encounter, and reflection to curriculum objectives.