If we don’t know how dexterous a robot is, how can we tell if it’s the right one for our task?

I was right to be skeptical. The UC Berkley team had done what many researchers have done in the past. They had invented an entirely new metric to measure dexterity. Their metric ignored the physical properties of the robot and instead focused on machine learning performance. There’s nothing inherently wrong with their new metric — which is really a measure of bin picking speed — but it certainly can’t be used to prove that the researcher’s robot is “the most dexterous robot ever created.” In order to say that, you’d need to measure all dexterous robots using the same metric.

This is not an isolated case. I’ve seen the same thing happen again and again. People use their own definition to try to prove a robotic system is dexterous.

There is no standard metric for dexterity — not until ISO nails it down at least. Unfortunately, that means you have to find your own way to assess a robot’s suitability for your task.

How to discover the dexterity needs of your task

The solution is to ignore the claims of “dexterity” made by manufacturers, the media, or researchers.

Instead, you should look closely at your robotic task and assess the specific performance that is vital for each of the steps. You can use this information to come up with your own definition of dexterity for your specific application.

Here are some of the important factors which relate to robot dexterity, along with questions that you can ask yourself to narrow down the needs for your task:

  1. Object size — How small are the objects that the robot will manipulate? Are there a variety of sizes or are all objects identical? How does this compare with the reach required of the robot?
  2. Object shape — What shape are the objects? Do they have many complex edges or a simple geometrical shape? Are they spherical or otherwise difficult to grasp?
  3. Gripping strategy — What are the different ways that the objects can be grasped (e.g. with an encompassing grip, internal grip, or suction)? Are there different ways to grasp the same objects? Are the objects delicate and so require a particular gripping strategy?
  4. Reachability — How much does the robot have to “stretch” to reach all important locations in the workspace? Does it need to use all the robot’s workspace, or just a small part of it? Does it need to approach locations from many different angles?
  5. Speed — What cycle time is required for each action?

We sometimes think that factors 1-3 are only related to the robot’s gripper and factors 4-5 relate to the manipulator. However, they are all inter-related. One factor in isolation does not necessarily define the robot as “dexterous” (e.g. a fast robot is not always dexterous), but together they give a picture of the dexterity needs of the task.

I should add that there are probably other factors besides those which I’ve listed which contribute to a robot’s dexterity, but these 5 are a good place to start.

Examples: High-dexterity tasks vs low-dexterity tasks

Let’s look how these factors relate to some fictional examples.

grippers for collaborative robots

Example 1: A high-dexterity assembly task

The task involves picking up several small, tough parts (1-3mm wide) with a variety of complex geometrical shapes. The robot needs to rotate the parts to align them and then assemble them together. Then, the robot must move 1.5m to the other side of its workspace to insert the assembled parts into a box. A cycle time of 30 seconds per part is required.

Dexterity factorRelationship to this task’s dexterity
Object sizeThe objects are very small, particularly considering the large reach required to the robot (1.5m).
Object shapeObjects are a variety of shapes which means that the gripper will need to be adaptable.
Gripping strategyThe shapes are geometric, so flexible grasping is likely to be needed. The objects are tough so don’t need delicate handling.
ReachabilityThe manipulation will be focused on two far-apart areas of the workspace. In the assembly area, it may need to be approach the objects from various angles.
SpeedThe required cycle time is very small for such a complex assembly application.

Most of the factors suggest a high degree of dexterity.

Example 2: A low-dexterity pick and place task

The task involves picking medium-sized, cuboid objects from one part of the workspace and placing them at an inexact location a short distance away. The objects are quite fragile and a cycle time of 30 seconds is required.

Dexterity factorRelationship to this task’s dexterity
Object sizeThe objects are not small and do not need to be placed precisely.
Object shapeThe objects are a uniform, basic shape which is easy to grasp.
Gripping strategyThe objects are a basic shape, but they are fragile so a force-limited or vacuum gripper may be necessary.
ReachabilityVery little of the robot’s workspace is required.
SpeedThe cycle time is low for the simple task and short distance required.

Most of the factors suggest a low degree of dexterity, despite the fact that the objects are fragile.

How to pick the best robot and gripper for you

As you can see, the dexterity required for a task can’t be defined by a single factor. Various properties of both the robot and its gripper will combine to determine the system’s dexterity.

For more news on robotics, AI and more, visit the Robotics & AI Channel.