Tight, twisty spaces difficult to detect?
A robotic mouse from BIT SQuRo A negative answer is given to this.
It can not only move flexibly in a small space, but also easily complete various movements and transformations.Such as squatting up, walking, crawling, etc.is simply an “artifact” to deal with sudden disasters or narrow pipelines:
Also within a small radius less than half of one’s own body lengthturn around quicklybite its own tail and make a 360° circle (the radius is much smaller than other robots):
Even strong enough to get up quickly after a fall.
Crucially, the mouse is still very capable of carrying loads—it has now successfully carried a weight that is 91% of its own weight (200 grams) through a field with a 20° inclination.
(Imagine the feeling of climbing up a hill with a bag about the same weight as yourself…)
The first author of the research paper, Professor Shi Qing of Beijing Institute of Technology, said that there are many legged robots on the market, but most of them are not good at coping with narrow spaces:
Large quadruped robots have strong transportation ability, but cannot enter narrow spaces; although micro quadruped robots can enter narrow spaces, their ability to carry heavy objects is limited.
The research results from the Beijing Institute of Technology have been published in IEEE journalssuperior.
After seeing the excellent agility and load capacity of this robotic mouse, let’s take a closer look!
Inspired by the rat who is not afraid of tight corners
Previously, few people have designed a small quadruped robot weighing less than 1 kg that can plan motion.Multimodal Control Framework.
Multi-modal control refers to a control method in which the strategy changes continuously with the operating state of the system. The most suitable control algorithm can be selected in real time, and the switch can be performed at the right time to make the system more stable, accurate and responsive.
Due to scale constraints, small robots have few hardware components, which leads to their low perception and processing power.
In addition, existing robotics research mainly focuses on dynamic stability and mechanical constraints, while ignoring the kinematic characteristics of a certain kind of robot-like creature.
Researchers found that mice are very agile in a variety of narrow and complex environments, so they preparedfrom a biological point of view“learning from the scriptures” in mice.
First, X-rays were used to record the skeletal structure of the mice in motion toExtract key kinematic jointsand then established a basic model of the quadruped robotic mouse.
The robotic mouse SQuRo weighs 220 grams, similar to the weight of an eight-week-old black-haired mouse; it is also about the same length as a real mouse.
The BIT team also endowed the robotic mouse with a multimodal motion planning and control framework, enabling it to perceive and process complex real-world environments.
Design the basic structure according to the 3 major abilities of mouse movement
According to X-ray analysis, the research team found that mice mainly rely on thisthree main functionsto combine various movements:
Therefore, the researchers configured the robot mouse with 12 degrees of freedom of movement (2 degrees of freedom in each limb, 2 degrees of freedom in flexion and extension at the waist, and 2 degrees of freedom in the neck), and 4 passive degrees of freedom to mimic joint flexion and extension. and turn.
Degrees of freedom are the number of independent variables. Specifically, if the total number of variables is N and the number of constraints is M, then the degrees of freedom F=NM.
The schematic diagram of the limb structure design of the robotic mouse is as follows:
▲ Figures a and b are the schematic diagram of the mechanism movement and skeleton model structure of the left forelimb respectively; c is the side view of the skeleton model of the left hind limb
The base of the hindlimb is a more curved rod than the forelimb to provide greater forward thrust—consistent with the fact that mice rely primarily on the hindlimb for thrust.
The researchers analyzed the behavior of the mouse and found that its turning movement was from the head to the torso, and then to the hips, gradually exerting force.
Benefiting from the flexible spine, the mouse can change direction quickly.
The cervical vertebra of a mouse is made up of several segments, and the rotation angle of the first cervical vertebra reflects the angle between the head and the torso.
In the joint rotation angle graph below, there are three peaks, corresponding to the three most obvious movements, namely: cervical flexion and extension,
Flexion and extension of the second thoracic vertebra of the forelimb, and flexion and extension of the hindlimb of the thirteenth thoracic vertebra.
Therefore, the researchers equipped the spine with three active degrees of freedom in flexion and extension for frontal turning movements in robotic mice.
Since neck rotation is rare in the daily activities of mice, the neck movements of real mice are of little significance for designing detection robots.
The researchers configured one active DOF for neck flexion and extension and one active DOF for neck adduction, both at the junction of the head and torso.
The robotic mouse has a total of 33 vertebral joints, and the researchers set the flexion and extension joint of the hind limb at the 22nd joint, which is similar to the corresponding joint position of the mouse.
Introduction of the research team
This research comes from Beijing Institute of Technology.
Shi Qing, the first author of the thesis, is currently a professor of Beijing Institute of Technology and the deputy director of the Institute of Intelligent Robots of the School of Mechanical and Electrical Engineering. Both undergraduate and doctoral degrees were graduated from Beijing Institute of Technology, and he did postdoctoral work at Waseda University. His main research directions are bionic robots and biomechanical integration.
This paper was jointly completed by Shi Qing’s tutor Huang Qiang, foreign academician Fukuda Minnan of the Chinese Academy of Sciences, and the bionic robot team led by Shi Qing.
The bionic mouse studied by the team was once evaluated by Janet Wiles, a computer professor at the University of Queensland, as “reaching the industry’s SOTA level”.
The team said that in the future, methods such as closed-loop control and in-depth dynamic analysis will be used to further improve the agility of the robotic mouse, and they are interested in commercializing it.
Where else do you think this robotic mouse could be used?
Paper address
.
[related_posts_by_tax taxonomies=”post_tag”]
The post BIT’s bionic electronic mouse will be used for post-disaster search and rescue: it is no problem to climb narrow pipes with heavy loads, and it can stand up when it falls. appeared first on Gamingsym.