robotswantdata · 615 points · 209 comments · il y a 20 heures · Open original
Comments
5 preview comments · loading full thread
Log in to use comments
Log in to h4cker, then connect Hacker News to publish comments.
LAlambdail y a 16 heures
Tried out the simulator and was surprised to find that the movement keys are ZQSD; then realized that's the equivalent of WASD on an AZERTY keyboard. Checked and sure enough, Pollen Robotics is a French company.
They may want to at least add a preference for keyboard layout, pretty sure that QWERTY and QWERTZ are much more common worldwide than AZERTY.
NUnum42il y a 4 heures
I found a few open-source small bipedal robots on the internet. There are actually many open-source robots available:
[1]. https://robotics.umich.edu/news/2024/a-smaller-simpler-biped...
https://github.com/daviddoo02/Legolas-an-open-source-biped
[2].https://github.com/MuShibo/Micro-Wheeled_leg-Robot
[3].https://github.com/Yuexuan9/Tinker
Quadruped:
[1].https://github.com/stanfordroboticsclub/StanfordQuadruped
[2].https://open-dynamic-robot-initiative.github.io/
MRmrandishil y a 9 heures
Page is so dense, couldn't find the specs:
Rockchip RK3566 processor with an AI accelerator, 1GB RAM, 32GB storage, Wi-Fi, Bluetooth, microphones, speaker, two NFC antennas and removable battery (~1 hour runtime). Onboard policy loop: 50 hertz, Dynamixel servos, weight: 800g.
Seven behaviors at ship: walking, sitting and standing, kicking, ground pickup, roller skating and self-recovery, can train additional behaviors locally or through Hugging Face Jobs, export to ONNX and deploy through open-source runtime. Software: SDK, a MuJoCo-based simulator, training scripts, tools for transferring policies from simulation to the physical robot.
PRprometheus1992il y a 15 heures
Not sure many people know this - most of the robotics stuff that you see in the news is based on MujoCo (Multi-Joint dynamics with Contact) engine, that is used primarily to create simulated environments in which these robots learn the RL policies. This project is maintained by Google deepmind - https://github.com/google-deepmind/mujoco
_J_joelil y a 18 heures
It's either this or https://mondorobotics.com/ and I'm torn. Obviously it's for my daughter and I will just be safety checking, definitely, 100%.
Comments
5 preview comments · loading full threadLog in to h4cker, then connect Hacker News to publish comments.
Tried out the simulator and was surprised to find that the movement keys are ZQSD; then realized that's the equivalent of WASD on an AZERTY keyboard. Checked and sure enough, Pollen Robotics is a French company. They may want to at least add a preference for keyboard layout, pretty sure that QWERTY and QWERTZ are much more common worldwide than AZERTY.
I found a few open-source small bipedal robots on the internet. There are actually many open-source robots available: [1]. https://robotics.umich.edu/news/2024/a-smaller-simpler-biped... https://github.com/daviddoo02/Legolas-an-open-source-biped [2].https://github.com/MuShibo/Micro-Wheeled_leg-Robot [3].https://github.com/Yuexuan9/Tinker Quadruped: [1].https://github.com/stanfordroboticsclub/StanfordQuadruped [2].https://open-dynamic-robot-initiative.github.io/
Page is so dense, couldn't find the specs: Rockchip RK3566 processor with an AI accelerator, 1GB RAM, 32GB storage, Wi-Fi, Bluetooth, microphones, speaker, two NFC antennas and removable battery (~1 hour runtime). Onboard policy loop: 50 hertz, Dynamixel servos, weight: 800g. Seven behaviors at ship: walking, sitting and standing, kicking, ground pickup, roller skating and self-recovery, can train additional behaviors locally or through Hugging Face Jobs, export to ONNX and deploy through open-source runtime. Software: SDK, a MuJoCo-based simulator, training scripts, tools for transferring policies from simulation to the physical robot.
Not sure many people know this - most of the robotics stuff that you see in the news is based on MujoCo (Multi-Joint dynamics with Contact) engine, that is used primarily to create simulated environments in which these robots learn the RL policies. This project is maintained by Google deepmind - https://github.com/google-deepmind/mujoco
It's either this or https://mondorobotics.com/ and I'm torn. Obviously it's for my daughter and I will just be safety checking, definitely, 100%.