Technical challenges and potential opportunities of dual-arm collaboration for household robot ALOHA

Recently, the Stanford-developed all-around housework Mobile ALOHA [1, 2] has sparked heated discussions online. In the video, ALOHA has two arms that can operate in coordination and can complete many complex tasks that require two-handed collaboration, such as cooking fried egg and shrimp, pouring water and taking out pills, hanging clothes on hangers, etc. These videos seem to indicate that robots are already able to replace humans in performing most tasks and have entered family life.

However, if we read the original paper carefully, we can find that most of the videos are not completed by the robot in autonomous mode, but by humans in remote control mode. In purely autonomous mode, the robot still makes various mistakes and cannot complete some complex tasks. These mistakes also show that the robot still needs further in-depth research and exploration when performing tasks, especially tasks that require the coordination of both arms.

Compared with the widely studied single-arm manipulation tasks, dual-arm manipulation tasks require the robot to coordinate and control the movement of two robotic arms to complete specific operations, such as assembly, disassembly, handling, and operating tools. Dual-arm collaborative operation can make the robot closer to human operation, achieve more efficient and accurate execution, and is also suitable for a wider range of application scenarios, such as food cooking, precision assembly, cleaning and other tasks.

So compared with single-arm manipulation tasks, what are the main difficulties in the research of dual-arm tasks? Among them, the most important difficulty lies in the ability of the two robotic arms to work together when handling the target object. In order to complete the specified task, the robot usually needs to control two robotic arms at the same time to operate different targets, and ensure that there is a high degree of coordination and cooperation between the target operations of the two robotic arms. For example, in an assembly task, when one hand grasps and fixes a part as a base, the other hand needs to grasp another part, rotate the part to the appropriate angle to be assembled, and finally install the part to the base part.

In this assembly task, we not only require the robot to learn the corresponding relationship between the two parts and the appropriate assembly angle, but also require the robot to accurately align the two parts and finally combine them into one. This requires a high degree of coordination between the two arms of the robot. Only when the two arms can cooperate and work together can the two parts be finally assembled and fit together at the right position and at the right angle.

On the basis of ensuring the collaborative working ability of the two arms when handling the target object, another difficulty lies in the path planning and motion control of the two arms during the task execution. The working space of each robotic arm is limited, and in order for the two robotic arms to collaborate to complete the task, the working spaces of the two robotic arms must overlap, so there is a possibility that the two robotic arms will collide or interfere with each other in the limited working space.

Path planning refers to the process of determining the motion trajectory of the robot arm from the starting state to the target state after the robot arm's starting posture and the target posture of the object being manipulated are given. Motion control refers to the process of controlling the robot arm to move along the planned path to ensure that the robot arm's movement is smooth and in line with expectations. With better path planning and motion control algorithms, we can better ensure that the two robots will not cross-interfere or collide during the execution of tasks and trajectory movement.

In addition, there are many other challenges in dual-arm tasks, such as higher precision requirements, because the operation error of the robot arm itself will be doubled in the collaborative operation of the two arms, and dual-arm tasks usually require more rigorous coordination, so the robot arm needs to achieve higher accuracy.

In addition, dual-arm tasks also require more efficient task allocation, because for some complex tasks, they need to be split and assigned to two robotic arms to ensure that they can work together effectively. Reasonable task allocation and coordination are essential to improving efficiency. Therefore, researchers still need to make more efforts before dual-arm robots can be truly used in daily homes.

Reviewing Editor: Huang Fei