EP.38 BUILD YOUR OWN AI-POWERED ROBOT

Mass production-ready human-like robots, table tennis playing robot by Google, burger delivery mobile robots & much more ...

Mass production-ready humanoids! 📊

Unitree has unveiled a production-ready version of its G1 humanoid robot, marking a significant step towards bringing human-like robots into everyday life. The G1 is an upgraded version of Unitree's H1 model, introduced just a year ago. Powered by an 8-core high-performance CPU, this robot boasts 23 degrees of freedom through powered joints in its arms, legs, and torso.

Its impressive physical capabilities include jumping, walking at speeds over 4.4mph, and navigating stairs even when littered with debris. One of the most notable improvements in the G1 is the replacement of the H1's arm stumps with three-fingered hands. These new appendages have demonstrated remarkable dexterity, capable of tasks ranging from soldering wires to flipping food in a frying pan. The robot's visual system, hidden in its LED-adorned head, combines a Livox Mid-360 lidar camera with an Intel RealSense D435 depth camera, enabling 3D perception of its environment.

While the G1's capabilities are impressive, potential buyers should temper their expectations. This robot isn't designed to be a plug-and-play domestic helper. Instead, it learns through imitation using Unitree's Robot Unified Large Model (UnifoLM). Its primary appeal lies in being an affordable platform for robotics research rather than an out-of-the-box household assistant.

As we await further details on the G1's release, it's clear that Unitree's latest offering represents a significant step towards making advanced humanoid robotics more accessible to researchers and enthusiasts alike.

---

Google created a robot that plays table tennis! 🏓

Google DeepMind has achieved a significant milestone in robotics by training a robot to play table tennis at a competitive amateur level. This breakthrough represents the first instance of a robot being taught to play a sport with humans at a human-comparable level, marking a major advancement in robotics and artificial intelligence.

The robotic system, consisting of an arm wielding a 3D-printed paddle, demonstrated impressive performance by winning 13 out of 29 games against human opponents of varying skill levels. While the robot excelled against beginners and amateur players, it still struggled against advanced competitors, highlighting both its capabilities and limitations.

The development process involved a two-part approach. Initially, the system was trained using computer simulations to master basic hitting skills. This was followed by fine-tuning with real-world data, allowing for continuous improvement. The robot's learning is facilitated by a dataset of table tennis ball states, including position, spin, and speed information. During matches, it collects performance data to refine its skills further, creating a feedback loop that enables it to adjust tactics and improve over time.

This research has implications beyond table tennis. It represents a step towards creating robots capable of performing complex tasks skillfully and safely in real-world environments like homes and warehouses.

Interestingly, human players reported enjoying their matches against the robotic arm, with even advanced competitors finding the experience engaging. Some participants even expressed interest in using the robot as a dynamic practice partner to improve their skills.

---

Burgers delivered by robots! 🍔

Serve Robotics, an autonomous sidewalk delivery company, has partnered with Shake Shack to offer robot-delivered meals through the Uber Eats platform. This collaboration marks a significant expansion of sidewalk robot deliveries on Uber Eats, which has been working with Serve in Los Angeles since 2022.

Customers ordering from select Shake Shack locations in Los Angeles through Uber Eats may now receive their food via Serve's autonomous robots. This partnership is part of Serve's ambitious plan to deploy 2,000 robots nationwide by 2025.

Shake Shack's senior vice president of digital experience, Steph So, expressed excitement about the collaboration, viewing it as an innovative way to enhance guest experiences both in and out of their restaurants.

This partnership represents a growing trend in the food delivery industry, showcasing the potential for autonomous robots to play a significant role in the future of urban food delivery services.

---

---

Meme of the week 🤖

So what should we call them? Credits to meme master Aaron Prather :)

---

---

Deep reinforcement learning for humanoids! 🦿

In this course, participants will explore deep reinforcement learning for walking robots using MATLAB and Simulink's Robotics System Toolbox™.

Upon completion, learners will be able to:

• Understanding the application of deep reinforcement learning (specifically DDPG algorithm) in controlling humanoid robot locomotion.

• Learning how to set up, train, and evaluate reinforcement learning using Simulink models in Simscape Multibody™ and Reinforcement Learning Toolbox™ environments.

• Selecting appropriate states, actions, and reward functions for the reinforcement learning problem.

Designing and implementing neural network structures for training the control policy.

---

Build your AI-powered robot! 🔨

Hugging Face, the open-source AI company has released a groundbreaking tutorial that guides developers through building and training their AI-powered robots. This initiative, based on their LeRobot platform launched in May, aims to democratize low-cost robotics and bring artificial intelligence into the physical world.

The comprehensive tutorial covers everything from sourcing parts to deploying AI models, making cutting-edge robotics technology accessible to developers of all skill levels. Remi Cadene, a principal research scientist at Hugging Face, describes it as a way to "unlock the power of end-to-end learning" in robotics, similar to how large language models work with text.

This initiative has significant implications for the future of AI and robotics. By democratizing access to these technologies, Hugging Face is fostering innovation across various sectors, from manufacturing to healthcare. However, it also raises important questions about the future of work, privacy, and ethical considerations of widespread automation.

As the technology matures, it has the potential to reshape industries and change how we interact with machines in our daily lives. For developers, entrepreneurs, and technical decision-makers, this tutorial represents a unique opportunity to engage with the future of robotics and AI.

---