Weekly Recap

Cool robots at ICRA 2024, GPT-4o, and Petoi Contest winners

ICRA 2024

The last week was a busy one in robotics. ICRA (International Conference on Robotics and Automation) 2024 was held in Yokohama, Japan. ICRA is a major gathering of both academics and exhibitors from robotic companies. It is geared more towards industrial robot design, so you would not find many papers on home robots. My favorite part of ICRA is watching all the cool robots from the exhibit area on Twitter (with hashtag #icra2024). As an example, this video shows the Boston Dynamic Spot in the Quadruped Robot Challange. And this video shows the robot parade. And if you wish to have a glimpse of humanoid robots, you can see this tweet. And some cool robots in a dance party here. There are lots more on Twitter and other social media. I would recommend you to check these out, some are fascinating.

If you were at ICRA 2024, and would like to help me out by posting your experiences here, please reach out to me. Use the button below.

---

Editor’s note: Thanks to all of you who have become paying subscribers. They really help fund the work here by helping me cover some costs: buying conference passes, robot components, late night pizzas, etc. If you want to angel invest in my work, please consider an annual membership offered at less than the cost of a dinner meal.

---

GPT-4o

Earlier this week OpenAI released GPT-4o (the ‘o’ standing for omni). GPT-4o is a multimodal model which can accept any combination of text, audio, image, and video and generates any combination of text, audio, and image outputs. The main innovation in GPT-4o is that it doesn’t require you to convert voice to text and then process, it can cosume voice (audio) natively, and return audio. Some of the example videos from OpenAI are very cool to watch. Currently, only the text version of GPT-4o is available for use, as per this tweet, we still need to wait for the voice version. I have always maintained that the domains of robotics and Large Language Models will intersect. With GPT-o, a robot can feed a video of its surroundings, and ask for an audio summarization. Its truly a revolutionary technology for robots designed for care.

Wirepod provides support for using GPT-4o in his latest commit (You will need to pull the latest wirepod code, and restart wirepod). We tested our beloved Vector robots with the latest Wirepod and generated this very cool video.

Petoi Contest Winners

Petoi announced the winners of the Petoi Robotics Challenge And Showcase Contests - Spring 2024 contest. If you are in to Petoi robots, you might find some of the videos cute. Our favourite video was Bittle visiting from Planet X. You can also check out our entry in the contest, we took Bittle to an animal farm and showed it to many animals.

Thanks again for being a subscriber.