Most robotic manipulators use a series of rigid structural elements connected by actuated joints, and it’s a good system—they can access a large workspace quickly and reliably. The downsides are that manipulators like these tend to be bulky and expensive, and they have single points of failure, such that if one of those actuated joints has an issue, the performance of the manipulator is seriously compromised.
Roboticists have tackled these problems in different ways. One is with snake-type robots, which use a continuous series of connected actuators for increased flexibility and hyper redundancy. These work great, except they’re still expensive, because they consist entirely of actuators. Another approach is using continuum robots, which rely on compliant actuation techniques (like cables or pneumatics) to achieve “infinite” degrees of freedom with lightweight and inexpensive hardware; the challenge with these, though, is that they tend to be difficult to control with precision, especially if your application needs rigidity.
At Ben-Gurion University of the Negev, in Israel, David Zarrouk and his students (Moshe Mann, Lior Damti, Gideon Tirosh, Liran Yehezkel, Otzar Dorani, Dar Zussman, and Dana Yael Erez) have developed a new kind of manipulator designed to offer the performance and redundancy of a snake robot, along with the low cost of a continuum robot. Their Minimally Actuated Serial Robot (MASR) is simple, lightweight, and modular, and offers plenty of advantages, as long as you’re not in any sort of hurry.
Essentially, MASR has similar capabilities to a traditional snake robot that uses a series of actuated joints, except that it can only move one joint at a time. This architecture makes this robot much, much slower than a snake robot, but it can also perform some neat tricks that give it a surprising amount of versatility:
The primary structure of the robot consists of an arbitrary number of passive links, which connect to each other at passive joints with locking clamps to keep them rigid. The only other defining feature of the links is that they include a rail for a movable actuator to travel along; at some point, the rail could also transmit power and data. The movable actuator itself uses just two motors: One to travel along the rail, and one to rotate the joints it stops on. No matter how many links you have, one movable actuator can handle them all—it may just take it a little while.
Besides this simplicity, the modularity and redundancy of this design are also very impressive. It’s easy to change the length by adding or subtracting links, and you can also add as many movable actuators as you need if you want to actuate multiple joints at once. It’s not just about making the design more flexible, either: Since it’s so easy to change things around, it means that if (or when) something breaks, fixing it becomes a lot easier.
It’s important to note that there are some things that the MASR is not very good at. Speed is one of them, of course, but so is moving an object along an arbitrary path. Unless the path happens to be an arc that lines up perfectly with one of the joints of the robot, moving something in (say) a straight line isn’t possible—rather, the MASR will spend a very long time repetitively actuating different joints and you’ll still just end up with a sort of zigzag. The figure below shows a simulated MASR trying to draw (appropriately enough) the letter Z.
You could mitigate this somewhat by adding more movable actuators, but if you’re looking for speed or precision of motion, this may not be the best robot for you. Fortunately, there are all sorts of applications where neither speed nor precision of motion are nearly as important as the end result.
For more details, we spoke with David Zarrouk via email.
IEEE Spectrum: How did you come up with this novel design?
David Zarrouk: We have been developing minimalistic robots for many years. We try to focus on minimizing the number of motors to increase the specific power (motor power divided by the total weight of the robot) of the robot. A few years ago, I realized that in many practical applications of serial robots we often have many motors that are located far from each other which we do not actuate them simultaneously. Since it is impossible to combine these motors together, I came up with the idea to move the motors from one place to another. This concept is simple but it is not bio-inspired by nature, since for living creatures it is impossible to move muscles from one place to another.
In what ways is this robot potentially more useful or capable than a continuum robot?
Continuum robots allow for practically infinite degrees of freedom (DOF). There is currently a lot of research going on continuous robots with amazing advances in the field. The robot which we propose here involves discretizing the robot into tiny segments, therefore reducing the number of DOF from infinity to a large number. However, this robot has some advantages.
First, unlike most continuum robots which are soft, this robot is kinematically speaking rigid. In the sense that it does not deform substantially under load. Second, the motion of the mobile motor along the links allows for new maneuvers that do not exist in other robots. For example, for and pick-and-place applications, transporting objects inside pipes, spraying (by placing the spray on the mobile motor and moving it along the links). Third, in this robot, the links and motors are totally decoupled. So it is relatively easy to change the number and size of the links and motors.
Why might this kind of robot be especially useful in space?
The number of satellites in space is increasing every year and many startup companies are racing to develop servicing satellites. Some of these space applications require long manipulators with a large number of DOF that can go around obstacles. As transporting loads to space is very expensive, our robot has an advantage as it has a low mass and weight. In addition, our robot has a low moment of inertia which is very beneficial especially in space applications since moving the arm would change the angular momentum and moment of inertia of the whole satellite causing it to rotate around itself. The lower inertia it has, the less control effort is required. The arm could be used to fix mechanical malfunctioning of the satellites, docking with other satellites, refueling satellites to increase their lifespan, or cleaning debris from space.
Can you describe how a robot like this could be modified to operate in 3D instead of 2D?
Currently our version is kinematically 2D. It makes the design, manufacturing, and control much easier. But we are planning to generalize it to 3D. A simple option is to add a horizontal actuated joint to the origin which would pitch the whole robot upwards and downwards. A second option will be to combine different joints whose axes are orthogonal to each other which will result in a full 3D maneuverability.
What would it take to commercialize a robot like this? And what are you working on next?
This robot design/concept is very simple and easy to operate. Therefore, we believe that it can be relatively easily transferred to different industries. We are very interested in collaborating with the industrial companies.
There are multiple research directions that we are currently pursuing. We are working on improving the design of the robot and improving the actuation of the robot with a reliable automatic mobile motor version. We are also developing locomotion planning algorithms for minimizing the time of the locomotion as a function of the number of mobile motors. In the future, the same concept may be applied to walking robots, in which the motors can change their position along the legs and even move from one leg to another when necessary.