Back to list

AI Products36氪

Geeks Build Cabbage-Priced Biped Robot for Under 3,000 Yuan, with Old Phones as Key Contributors

36氪1 min read
Read original
Geeks Build Cabbage-Priced Biped Robot for Under 3,000 Yuan, with Old Phones as Key Contributors

AI Summary

UCLA's RoMeLa lab has developed PhoneBot, a 1.8 kg biped robot with 13 degrees of freedom and a materials cost of about $400 (roughly 2,677 yuan). The robot is powered by a smartphone that serves as its sensor suite, onboard computer, and interface, and by using used older phones, total cost stays under 3,000 yuan.

AI Insights

PhoneBot demonstrates that a capable bipedal robot can be built for under $500 by repurposing used smartphones as sensors, compute, and interface, pointing toward a 'cheap body plus swappable phone brain' model similar to the Raspberry Pi ecosystem. This could lower barriers for universities, startups, and hobbyist developers, accelerating open-source embodied AI research. However, limited onboard compute, reliance on external PCs for speech processing, and the absence of arms mean the near-term impact is more likely in education and R&D than mass consumer markets.

Analysis written by AI from the original article. Check it against the source.

Translated text

UCLA's RoMeLa lab recently introduced PhoneBot, a biped robot 483 mm tall, weighing 1.8 kg, with 13 degrees of freedom. Each joint uses a servo that costs $23.9. A smartphone is fixed to the torso and serves as the sensor suite, onboard computer, and user interface. The robot can walk, stand up, follow people, take photos, and hold conversations. Testing used older models such as the Moto G2025 and Samsung Galaxy A15, which sell secondhand for about $100 or less. Even including the phone, total materials cost is under $500, or roughly 3,000 yuan. Motion control works as follows: an Android app reads the phone's gyroscope and accelerometer along with the current angles of the 13 motors, and a pre-trained small model computes the target position of each leg, then sends commands over Wi-Fi to a small converter board that drives the motors. Walking, standing up, and turning are all handled this way. Image recognition and localization run entirely on the phone, using the rear camera and sensors. Speech-to-text and large-model inference are handled by a laptop placed nearby. CPU load on the phone is about 6-10% for gait calculation, 16-18% for person and object recognition, and 25-34% for localization. PhoneBot is an experimental platform with no arms, no payload capacity, and no stair-climbing ability. Still, treating old phones as replaceable robot brains with a durable body could lower barriers for developers and create a modular model. As on-device AI chips improve, the reliance on external PCs for speech processing may also decline.

Related News

3

AI Products36氪

Alibaba, ByteDance and Meta Ignite Personal Agent Race: No Table Without a Good Model

In late September, Alibaba announced at its Yunqi Conference that Qwen (Tongyi Qianwen) would accelerate personal agent development, and on October 13 it unveiled 'Qwen Go!', a personal assistant in limited invitation testing. ByteDance (Doubao), Tencent and others have also launched personal agent projects, sparked by Meta's Muse. On the model side, Meta's Muse Wonder and Alibaba's QwenGyre reinforcement-learning framework aim to strengthen long-running agent capabilities.

AI Products36氪

Apple Unveils 12 Home Products at Once, Including HomeView; Siri AI to Control the Whole House

Apple last night unveiled four home product lines: the HomeView smart home hub with a screen, the latest Apple TV 4K, the new HomePod mini 2, and the Home Collection smart home accessories. HomeView and Apple TV 4K both feature Siri AI, and the HomePod mini received its first update in six years. In mainland China, the HomePod mini 2 goes on sale on October 30 for 1,099 yuan.

AI Products36氪

Tech Giants at Home and Abroad Scramble for the Muse Entrance, But the Personal Agent Remains a False Proposition

A month after Meta launched the personal agent Muse, touted as the most captivating product since ChatGPT, ByteDance, Tencent, Alibaba, and Ant Group have rushed out similar offerings. However, Muse's daily net new users fell 62.4% from the September peak, and analysis suggests the model struggles to fit the domestic Chinese market in both product logic and business model.