humanoid.observer
44 robots on sale today

Berkeley Humanoid

Berkeley Humanoid is a mid sized, low cost bipedal research robot from UC Berkeley's Hybrid Robotics Lab, presented in a July 2024 paper as a reliable platform for learning based control. It was designed to be built and repaired by a small team and to survive many falls. It led to the fully open source Berkeley Humanoid Lite in 2025.

Research not for sale Score 22/100, niche Generation 1 Succeeded by Berkeley Humanoid Lite Checked 2026-09-04
No written verdict yet

We have not sat down with this one. The score below is computed from the data we hold, and the facts on this page are sourced. It is not on sale, so there is nothing to recommend.


22/100
Niche
Capability 13/30
Value 0/25
Buyability 2/20
Ecosystem 7/15
Proof 0/10
Computed from the data · how the score works

How the score is built

Capability is what the machine physically does. Value compares it with others at a similar price, not with the whole market. Buyability asks whether you can really order one. Ecosystem is the software and community around it. Proof is what it has done in the field, which is where most humanoids score badly and should.

Read the full method

Instead of this

What people buy instead.

Compare these
Status
Research
Category
research
First revealed
July 2024
Form
bipedal
Height
85 cm
Weight
16 kg
Payload
n/a
Degrees of freedom
12
Battery
n/a
Runtime
n/a
Walking speed
n/a
Actuation
electric, custom quasi direct drive actuators
Units deployed (est.)
not disclosed
Teleoperation
no

Where it stands

Research platform of the Hybrid Robotics Lab; not for sale, succeeded by the open source Berkeley Humanoid Lite.

What stands out

  • Designed for reinforcement learning research
  • Low cost quasi direct drive actuators
  • Withstands repeated falls

Autonomy, observed

Paper and videos show learned walking on varied terrain and robustness to pushes, tested in the lab and outdoors on campus

Hands, sensors, brain

  • Hands: none in the base version
  • Sensors: IMU, joint encoders
  • Compute: onboard computer running learned policies
  • AI stack: reinforcement learning policies trained in simulation

Price and volume

reported build cost under 10,000 dollars per unit

In the blog

Berkeley Humanoid is covered in these 2 articles

All articles
Same maker

Other robots by University of California, Berkeley.

Sources

  1. 01
  2. 02

Spotted a wrong price or a broken fact? Send a correction. Fixed within 24 hours when sourced.