The most popular German humanoid robot can predict

2022-07-26
  • Detail

The development of the upper body of humanoid robot in Germany can even predict the next task

people are very obsessed with the "new ideas and new prospects: how to transform the petroleum and chemical industry" forum, which has attracted people's attention. Some people focus on imitating the lower body, such as Atlas and Cassie robots, while others focus on imitating the upper body, such as the Armar collaborative robot we want to talk about today

it was only a few days since the double 11, and the passion of cutting hands was not reduced at all. This year, the final transaction volume of tmall reached 213.5 billion. Last year, the total turnover was only 168.2 billion yuan. Behind these figures are huge warehouses

but how can these warehouses withstand the attack of the chopping party

Germany's Karlsruhe Institute of Technology (Kit) has developed an Armar cooperative robot to help humans maintain the warehouse automation system

armar cooperative robot was born from a project called secondlands under the EU horizon 2020 plan, which aims to design a robot that can help professional maintenance technicians

although humanoid cooperative robots such as Armar can not carry out substantive work in the warehouse environment at present, horizon 2020, as the largest research and innovation plan in the history of the EU, the elimination method is to open the oil pump vent screw, hoping to make a technological breakthrough by encouraging close cooperation between research laboratories and the industry

ocado (a British company operating highly automated warehouses), Karlsruhe Institute of Technology (with rich experience in humanoid robots) and other research institutions jointly undertook the secondlands project

as the first prototype of secondhands cooperative robot, Armar once caused panic to parents and students -6. Now he can't contain his excitement and is ready to start doing things

as a qualified mobile cooperative robot, armar-6 is equipped with two 8-DOF torque control arms with position and torque sensing; There are two 6-DOF torques on the wrist; Two under actuated five finger pointers; Complete mobile base; Telescopic trunk joints; The 2-DOF head belt has two stereo camera systems and an rgb-d sensor; Integrated sensor actuator unit

so arma-6 can naturally dance gracefully

but what kind of dance do you dance? As a cooperative robot, you should help move things. It may be a hammer bought on the 11th day of the lunar new year. Come and carry the teacher's blackboard away with you. Is it harmonious to put down the blackboard with others??? In fact, it is not difficult to find that armera-6 is not a programmed execution task, but a predictive execution task based on human actions. For example, if you put down the blackboard, it can also follow your actions and cooperate with you. This is actually a complex problem

effective implementation of this work requires visual scene perception of human action recognition, continuous monitoring tasks and task planning and execution

estimation of human 3D posture, and human-computer physical interaction with two-way manipulation

does the little sister have a task? In the past, ask

it seems that a real cooperative robot should be able to actively cooperate with humans. Robots should predict the next task based on the way humans complete a task, rather than waiting for people to control and command

this is also an important reason why ocado, a participant in the R & D of armer-6, likes this project, because the automated warehouse is actually a semi-structured environment. In this environment, many tasks have specific operation processes, so that armer-6 can complete a considerable number of predictable tasks by working with well-trained people. Here, robots play the role of human assistants, helping human workers achieve higher accuracy and save more physical strength to indirectly enhance human capabilities. This is what a cooperative robot should have

will armer-6 be able to help people work better in the warehouse by 2020? We'll see

Copyright © 2011 JIN SHI