Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

I am really impressed by how easy Spectacular AI makes visual-inertial odometry with an OAK camera. Just "pip install spectacularAI" and run a Python script. Running VIO/SLAM used to be much harder in my opinion.

Yu Xiang

7,185 subscribers

70,818 views • 9 months ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

26 Comments

Marco Franzon9 months ago

Really impressive! Perfect for drones

r3333d9 months ago

I'm actually researching monocular SLAM right now. I want to run in real-time from a drone video stream.

Yu Xiang9 months ago

You can try the spectacularAI package if you have an OAK camera, very easy to run

MechatronicsEngineer9 months ago

Crazy how much simpler it’s getting did you tweak anything, or just ran it as is?

Yu Xiang9 months ago

Plug the camera and run it. Because they already have support for these cameras:

payback9 months ago

spectacular ai doesn't work on arm , hence useless for embedded applications, also it's a proprietary sdk, u can do vio slam with native depthai sdk

Danilo Gasques, Ph.D.9 months ago

This looks so good

AIトレンド速報｜最新ニュース & 活用術9 months ago

Yeah, the pip install + script approach really simplifies VIO/SLAM with OAK. Huge improvement!

Hongyu Li9 months ago

How's their depth quality?

Yu Xiang9 months ago

It is active vision using IR. So the depth quality is not that great. The plan is to get depth from FoundationStereo

Arash Tajik9 months ago

curious how your experience was with different cameras; I saw you have been trying OAK and other brands in previous posts.

Yu Xiang9 months ago

I am trying to find a RGB-D/stereo camera for ego-centric data. So far, I decided to use OAK-D Pro W for its wider FOV. Overall, ZED and OAK cameras have more options.

Andrey Tatarinov9 months ago

Curious, can it work on a monocular video?

Yu Xiang9 months ago

Yes, it can be used with one camera

Sir Kabuchanga9 months ago

@OpenRoboticsOrg Wow interesting, am trying to imagine a the applicants.

Tyler Gibson9 months ago

Would be cool for 360 cams

nyx9 months ago

the gap between "theoretically possible" and "actually usable by normal humans" is where most good tools die. nice to see one that made it across

Ashley Feng9 months ago

sounds super cool, i love when tech gets easier to use

Jeffery Wang9 months ago

How's position track quality, and drift performance? using the native free spectacular AI pack?

Yu Xiang9 months ago

Looks pretty good to me

Jeffery Wang9 months ago

Nice, Thanks !

Leo Chien9 months ago

How fast the odometry data can output with the camera?

luke9 months ago

Does it support ZED cameras? I'm a big fan of ZED 2i

United Records9 months ago

wow

setoFPV4 months ago

can we use T265 with Spectacular AI libs?

Himanshu Kumar9 months ago

I'm impressed; Spectacular AI's visual-inertial odometry is easy with an OAK camera.

Related Videos

Easiest way to calibrate a camera on Linux. No ROS, no Docker, no CMake. Just FFmpeg and Spectacular AI command line tool installable via Python pip: sudo apt install ffmpeg pip install spectacularAI[full] sai-cli calibrate your-data.mkv

Easiest way to calibrate a camera on Linux. No ROS, no Docker, no CMake. Just FFmpeg and Spectacular AI command line tool installable via Python pip: sudo apt install ffmpeg pip install spectacularAI[full] sai-cli calibrate your-data.mkv

Otto Seiskari

92,489 views • 1 year ago

Built an open-source Jupyter Notebook that brings the power of OpenAI's code interpreter into your local Python development environment. Watch how easy it is to use Thread to edit code I wrote, chat with my notebook, or generate charts just using prompts. We built Thread using Next.js, Chakra and Project Jupyter and you can install it by running: pip install thread-dev

Built an open-source Jupyter Notebook that brings the power of OpenAI's code interpreter into your local Python development environment. Watch how easy it is to use Thread to edit code I wrote, chat with my notebook, or generate charts just using prompts. We built Thread using Next.js, Chakra and Project Jupyter and you can install it by running: pip install thread-dev

Ali Shobeiri

112,687 views • 2 years ago

lite visual inertial odometry for edge devices... getting there!

lite visual inertial odometry for edge devices... getting there!

ud

15,193 views • 7 months ago

Open Interpreter's new release looks great for locally running LLMs The tool lets LLMs run code (Python, Javascript, Shell, and more) locally. You can chat with Open Interpreter through a ChatGPT-like interface in your terminal by running $ interpreter after installing. Just do `pip install open-interpreter` and then just run `- interpreter --local` sets up fast, local LLMs. Congrats killian 👏

Open Interpreter's new release looks great for locally running LLMs The tool lets LLMs run code (Python, Javascript, Shell, and more) locally. You can chat with Open Interpreter through a ChatGPT-like interface in your terminal by running $ interpreter after installing. Just do `pip install open-interpreter` and then just run `- interpreter --local` sets up fast, local LLMs. Congrats killian 👏

Rohan Paul

154,697 views • 2 years ago

You should really stop using Zapier, Make or n8n Generating a Python script and running it is much faster (and cheaper) using: - o3-mini / Claude / Gemini to create the script - Python Code Runner (free Chrome extension) to run it And it only takes a few minutes. Link below

You should really stop using Zapier, Make or n8n Generating a Python script and running it is much faster (and cheaper) using: - o3-mini / Claude / Gemini to create the script - Python Code Runner (free Chrome extension) to run it And it only takes a few minutes. Link below

Paul Couvert

166,508 views • 1 year ago

📽️ I just used AI to make a video explaining generative AI. What is it? How will it be used? AI script written by OpenAI Person in video generated by Synthesia 🎥 Background b-roll created by @pictoryAI Combined everything with Final Cut Pro Took a few minutes total!

📽️ I just used AI to make a video explaining generative AI. What is it? How will it be used? AI script written by OpenAI Person in video generated by Synthesia 🎥 Background b-roll created by @pictoryAI Combined everything with Final Cut Pro Took a few minutes total!

Matt Schlicht

1,136,971 views • 3 years ago

I am very impressed with Hunyuan3D-2GP image to 3d model! Gave it the sneaker test and I used the lowest settings possible,will have to start experimenting to see how much I can push my GPU but even this is a really great base.

I am very impressed with Hunyuan3D-2GP image to 3d model! Gave it the sneaker test and I used the lowest settings possible,will have to start experimenting to see how much I can push my GPU but even this is a really great base.

DreamStarter

36,655 views • 1 year ago

Kuro really outdoing themselves with the visual storytelling every arc.. how can I not be impressed man

Kuro really outdoing themselves with the visual storytelling every arc.. how can I not be impressed man

Ozzy 🇳🇬

10,577 views • 2 months ago

Really impressed with the new video feature in @NotebookLM. I asked for a history of Disneyland. It kicked off a deep search, pulled 58 sources, used them to write a script, and generated a 5 minute video with a mix of real and AI visuals. How it works + a clip 👇

Really impressed with the new video feature in @NotebookLM. I asked for a history of Disneyland. It kicked off a deep search, pulled 58 sources, used them to write a script, and generated a 5 minute video with a mix of real and AI visuals. How it works + a clip 👇

Justine Moore

60,109 views • 7 months ago

First demo of our new aerial Visual-Inertial Positioning System (VIO+VPS): 40m positioning accuracy (CEP) in a 50 kilometer flight using a camera, consumer-grade IMU and a barometer! No GPS

First demo of our new aerial Visual-Inertial Positioning System (VIO+VPS): 40m positioning accuracy (CEP) in a 50 kilometer flight using a camera, consumer-grade IMU and a barometer! No GPS

Otto Seiskari

101,767 views • 2 years ago

Building AI agents that handle real money is hard. Verifying their work is harder. We just made it easy. Today, we're releasing the ChaosChain SDK: a one-line install to give your agents the world's first Triple-Verified Stack. pip install chaoschain-sdk

Building AI agents that handle real money is hard. Verifying their work is harder. We just made it easy. Today, we're releasing the ChaosChain SDK: a one-line install to give your agents the world's first Triple-Verified Stack. pip install chaoschain-sdk

Sumeet Chougule

19,544 views • 1 year ago

Introducing ClawApp! 🦞 The first open-source desktop app that makes @OpenClaw easy to install, run, and manage. No scripts. No fragile setup. Just local agents, up and running in minutes. Download it now at

Introducing ClawApp! 🦞 The first open-source desktop app that makes @OpenClaw easy to install, run, and manage. No scripts. No fragile setup. Just local agents, up and running in minutes. Download it now at

Sahara AI 🔆

22,720 views • 8 months ago

Want to run Claude Code in yolo mode but don't want it to potentially cripple your system? Easy 🔥 Install Docker to run it in a sandbox w/ this script: curl -sSL | bash Script below:

Want to run Claude Code in yolo mode but don't want it to potentially cripple your system? Easy 🔥 Install Docker to run it in a sandbox w/ this script: curl -sSL | bash Script below:

Ian Nuttall

48,219 views • 1 year ago

As part of my exo ego pipeline, I haven't yet found a VIO/Slam solution that hits all of the features I want. Now that agents make it super easy to write low level systems code, I've almost finished building a portable rust/python slam pipeline that works on both a 5090 AND a Rockchip-class computer. To really make this work, I needed a good ground truth, and the wonderful work from Collabora and Mateo de Mayo on the monado slam dataset has been indispensable. I registered it as a Rerun dataset in the OSS catalog to make it very easy to verify that the GT actually looks right and use it as an eval, and it has worked REALLY well. I'll have more on the VIO pipeline here in the coming days

As part of my exo ego pipeline, I haven't yet found a VIO/Slam solution that hits all of the features I want. Now that agents make it super easy to write low level systems code, I've almost finished building a portable rust/python slam pipeline that works on both a 5090 AND a Rockchip-class computer. To really make this work, I needed a good ground truth, and the wonderful work from Collabora and Mateo de Mayo on the monado slam dataset has been indispensable. I registered it as a Rerun dataset in the OSS catalog to make it very easy to verify that the GT actually looks right and use it as an eval, and it has worked REALLY well. I'll have more on the VIO pipeline here in the coming days

Pablo Vela

20,588 views • 1 month ago

Can an inexpensive, off-the-shelf IMU be the only sensor to estimate the full state (position, velocity, orientation) of a quadrotor flying through a track at high speed and even be on-pair with vision-based localization? The answer is yes, within certain limitations! In this #RAL2023 paper, we propose a learning-based odometry algorithm that couples a model-based filter driven by the inertial measurements with a learning-based module with access to the control commands. Our system outperforms by a large margin the state-of-the-art visual-inertial odometry (#VIO) algorithms and the state-of-the-art learned-inertial odometry algorithm, #TLIO, for the task of drone racing. Additionally, we show that our system is as accurate as a VIO algorithm that uses a camera to localize to a known map of the racing track. The main limitation of our approach is that it cannot generalize to trajectories that have not been seen at training time. However, in drone racing competitions, the track is known beforehand. Human pilots spend hours or even days of practice on the race track before the competition. Similarly, our system can be trained with the data collected during practice time and deployed during the competition. Future work will investigate how to generalize to trajectories not seen at training time. The code is released! Paper: Video: Code: Kudos to Giovanni Cioffi Leonard Bauersfeld Elia Kaufmann European Research Council (ERC) University of Zurich UZH Science UZH Space Hub NCCR Robotics Aerial Core #RAL2023 #IROS2023 #SLAM

Can an inexpensive, off-the-shelf IMU be the only sensor to estimate the full state (position, velocity, orientation) of a quadrotor flying through a track at high speed and even be on-pair with vision-based localization? The answer is yes, within certain limitations! In this #RAL2023 paper, we propose a learning-based odometry algorithm that couples a model-based filter driven by the inertial measurements with a learning-based module with access to the control commands. Our system outperforms by a large margin the state-of-the-art visual-inertial odometry (#VIO) algorithms and the state-of-the-art learned-inertial odometry algorithm, #TLIO, for the task of drone racing. Additionally, we show that our system is as accurate as a VIO algorithm that uses a camera to localize to a known map of the racing track. The main limitation of our approach is that it cannot generalize to trajectories that have not been seen at training time. However, in drone racing competitions, the track is known beforehand. Human pilots spend hours or even days of practice on the race track before the competition. Similarly, our system can be trained with the data collected during practice time and deployed during the competition. Future work will investigate how to generalize to trajectories not seen at training time. The code is released! Paper: Video: Code: Kudos to Giovanni Cioffi Leonard Bauersfeld Elia Kaufmann European Research Council (ERC) University of Zurich UZH Science UZH Space Hub NCCR Robotics Aerial Core #RAL2023 #IROS2023 #SLAM

Davide Scaramuzza

37,061 views • 3 years ago

Such a good skill for visual explainers. All SVG and animated. Prompt I used: install skill /anatomy Explain how Seedream 5.0 is able to split an image into layers and what cool things you can produce with it. Do that in a single HTML file. Change the camera/zoom in for scene 1 to 5 to explain more clearly.

Such a good skill for visual explainers. All SVG and animated. Prompt I used: install skill /anatomy Explain how Seedream 5.0 is able to split an image into layers and what cool things you can produce with it. Do that in a single HTML file. Change the camera/zoom in for scene 1 to 5 to explain more clearly.

Meng To

113,786 views • 19 days ago

I am an American citizen, proud to be an American citizen, who loves my country, and loves India so much. But this is so true, and accurate. India is running. Yes, India is running high. Just a brilliant analysis Palki Sharma

I am an American citizen, proud to be an American citizen, who loves my country, and loves India so much. But this is so true, and accurate. India is running. Yes, India is running high. Just a brilliant analysis Palki Sharma

Renee Lynn

567,588 views • 1 year ago

As time goes on, I find it harder and harder to suppress the pain I feel when I think of how much better things used to be.

As time goes on, I find it harder and harder to suppress the pain I feel when I think of how much better things used to be.

Vinnie Sullivan

3,731,969 views • 10 months ago