Unitree AlienGo

• Depth perception vision system
• High explosive sports performance
• Adaptive complex terrain
• Long battery life & heavy load
• Rich external interface
• All-in-one body
• Advanced protection level
• Built-in intelligent air cooling system

Super-body Long battery life
The patent-pending design of the whole machine stems from the self-developed technology. The weight and performance consumption are calculated to the extreme, bringing a long life experience.

The self-developed power lithium battery pack makes it have excellent battery life. The maximum operating time can reach 4.5 hours. The built-in intelligent air cooling system can automatically sense the temperature. The body is made of environmentally friendly materials such as polymer plastic, carbon fiber and aviation aluminum. , Passed the relevant product environmental safety certification, the protection level is extremely high, the fuselage is integrated design, all the power supply, signal line built-in package design

Compound control of joints by force control technology
It can realize full control of 3-axis posture and position, so it has strong multi-terrain adaptability and can run stably on rugged gravel roads and grass roads.

Good stability
Even if there is external force interference, the attitude can be adjusted quickly to achieve stable operation.

Excellent athletic performance

• 12 high-performance servo motors

Special movements such as fast running, backward running, side-to-side movement, in-situ turning, creeping forward, rollover, jumping, backflip, obstacle crossing, obstacle avoidance, climbing in place after falling to the ground, up and down slopes/steps, etc.

The maximum walking speed exceeds 1.5m/s

The fuselage has a good ability to resist impact. When running, jumping, colliding with foreign objects or falling, the fuselage can well resist the impact load.

Good openness
Software control interface is divided into high-level interface and bottom interface
Control interface supports C/C++, ROS, etc.
Allow reading various sensor data
Allows control of single motor or single leg movement

Rich external interface

Developer version with onboard PC and open corresponding interface
Allow users to carry their own modules (such as vision camera, lidar, robotic arm, GPS system, etc.) for interactive control
External interface includes: HDMI x 2, Ethernet port x 2, USB 3.0 x 2, USB 2.0 x 1, 485 port x 1

Depth Perception Vision

• Depth camera | 2 SET
  Global shutter and wide field of view
  The minimum sensing depth is about 0.11 m
  Up to 1280 x 720 depth resolution

• Visual odometer camera | 1SET
  Highly optimized V-SLAM, closed-loop offset is less than 1%, and the delay between gesture motion and motion reflection is less than 6 ms.
  The fisheye lens imager, combined with a near hemispherical field of view of 163±5 ° , enables fast tracking with fast movement.

Depth Vision -3d Map Real-time
Creation and Navigation Planning

3D Environment Construction
In the process of motion, the robot uses the cameras to obtain the color and depth information of the environment, and then reconstructs the 3D spatial information of the object with the help of a specific vision algorithm.

Probability Map
Octomap(probability map) are built by using cameras that detect the robot's surroundings as it moves provide back obstacle data.

Dynamic Obstacle Perception
When the robot encounters a dynamic obstacle, it will refresh the current map data within a certain range, thus discarding the "moving artifact" left by the dynamic obstacle on the map.

The Global Positioning
During the process of map creation,the global and local real-time positioning functions are available. The map will follow the camera's perspective in real time, and support real-time zooming in, zooming out, moving and arbitrary rotation.

Loop Detection
The robot can maintain a high loop-back accuracy in a wide range of fields, a high positioning accuracy within a certain range, and can maintain stability within a certain oscillation amplitude, with drift or loss.

Human Posture Recognition Tracking and Face Recognition

• Body Posture Recognition
  The color camera can identify the specific posture of the person according to the deep learning model, and conduct human-machine interaction. The robot can make corresponding movements according to different body postures.

• Human Skeleton Perception
  The robot can analyze and calculate the two-dimensional skeleton information of the human body according to the color information from the perspective, and further analyze and calculate the three-dimensional skeleton information and motion information of a specific character using depths of field.

• Target Person Tracking
  When there is more than one person in the scene, someone can tell the robot to lock he/she by a certain posture (for example, raising the left hand). Thereafter, the robot will follow the movement of the target, even during the movement.