NASA Datanauts Water Cooler Chat: Autonomous Design of Modular Robots

Editor's Notes

• #4: Such as industrial robots that are used for welding or service robots that are used for production and distribution of goods. Robots also can be employed for jobs that are too dangerous or dull for humans; such as space exploration, nuclear sites inspection, and hazardous rescue operations. But, these robotic applications are not widespread because of their high cost and long development time. Moreover, the pre-designed elaborate structure of robots is not sufficient to perform all tasks.
• #5: Therefore, modular robots are used to overcome the limitations of developing mainstream robots by offering a method of rapid and cost-effective production. Modular robots are efficient because they are flexible. They can form multiple shapes that can perform different tasks, while the mainstream robots are designed to perform a specific task. Designing modular robotic systems to accomplish a variety of tasks, sometimes in dynamic or unstructured environments, can be very challenging for human designers, especially with the lack of practical design guidelines. If you look at this robot: It has a head, arms, and legs. It resembles what humans are used to see and think of. Therefore, there is a need for an automated design system for modular robots to save time, effort, and think out of the box to come up with efficient task-based robotic design that humans can’t think of.
• #6: Evolutionary computation is employed to design modular robots autonomously.
• #8: In Self-Assembly: when the modules mix, they can bind to form the goal robotic design. Self-Reconfiguration allows a robot to change its structure from one design to another to form a new shape and perform a different task. Self-Repair robots have 2 qualities: The redundancy of modules they carry allow the robot to continue performing the task at hand even after a failure. The robot can detect the failed module, eject it and continue achieving the task with less efficiency or replace it with a fresh one to keep the system running with full efficiency. Self-replicating machine is a type of autonomous robot that is capable of reproducing itself autonomously using raw materials found in the environment
• #9: Another motivation is space exploration, because modular robots are smaller in size and can accomplish several tasks, the y can be sent to the space to assist the astronauts or they could be sent in a purely robotic mission to manipulate solar panels or it can have cameras for space exploration. Modular robots can be used for search and rescue applications in disaster areas. Modular robots have security and surveillance applications such as under vehicle inspection, wide-area perimeter surveillance, and high value asset monitoring.
• #10: 2017 Dtto is a modular self-reconfigurable robot that utilizes 3D printing technology to print the cases of the modules. The electronic parts can be found on ebay for less than 55 dollars. Dtto is capable of performing different locomotion modes. Each module is equipped with sensors, rechargeable batteries, and an Arduino nano for computational purposes. The reason I decided to use Dtto is that: I did experiments using Cubelets and was not satisfied with the results, because the predetermined structure has many limitations which won’t produce the variety of results we are looking for. When I reviewed the research to find the best performing modular robot… It was M-TRAN and ATRON. ATRON is spherical and you must have a meta-module that is composed of 3 modules as the seed… This will increase the complexity. The winner was M-TRAN and by the time I found out that result. Dtto came out as an economic alternative to M-TRAN where you can build and customize the modules by adding different sensors.
• #12: Dtto is composed of double-cube modules inspired by M-TRAN III modular design. Every module comprises two identical semi-cylindrical boxes connected by a link, such that each box can rotate 180˚ independently around that link; The single Dtto module has 6 connection faces; thus each box has 3 connection faces of opposite polarities – one block is active (male) while the other one is passive (female). Each module includes actuators, sensors, micro-processors, and electro-magnetic connectors.
• #14: In order to represent a modular robot using AIM, we first need to describe the structural topology of that robot using a labeled kinematic graph, where each vertex represents a link module and each edge represents how two vertices are connected. The incidence matrix of a modular robot composed of n modules is composed of n rows (vertices) and n-1 columns (edges). Every entry of 1 demonstrates a connection between the link modules represented by the vertices. Zero entries illustrate that there is no connection. Then, the AIM of that graph is obtained by replacing every entry of one in the incidence matrix of the graph by the label of the associated connecting socket and keeping the zero entries unchanged.
• #16: In order to facilitate autonomous robotic assembly and reconfiguration, a kinematic model is proposed to express modular parameters of the robot; i.e. the type and sequence of the modules. In this kinematic model, each module has a designated body coordinate system originating from the center of that module. The z-axis is perpendicular to the link connecting the semi-cylindrical boxes of the link module and the x and y axes are pointing towards the connection sockets. This coordinate system is useful for labeling the faces of the link modules, because the orientation of every connected link modules in a robotic structure can affect the resulting locomotion gait of that robot. The labeling of the faces starts with 1 goes on until 6
• #17: AIM must be modified in order to include more information that can differentiate between the two robots. In this research, AIM is extended by adding to the original AIM, the orientation of the module link in the 3D world relative to its own frame; such that each edge is described by the label of the connection socket followed by the rotation angles around the x, y, and z axis. The rotation angles could have the values of 0˚, 90˚,180˚, or 270˚ and the rotation is performed relative to the frame of the module in the clockwise direction.
• #18: AIM must be modified in order to include more information that can differentiate between the two robots. In this research, AIM is extended by adding to the original AIM, the orientation of the module link in the 3D world relative to its own frame; such that each edge is described by the label of the connection socket followed by the rotation angles around the x, y, and z axis. The rotation angles could have the values of 0˚, 90˚,180˚, or 270˚ and the rotation is performed relative to the frame of the module in the clockwise direction.