Back to Past Projects

Soft Grippers and Bugbots

Aaron M. Dollar, Robert D. Howe

Support provided by the Office of Naval Research

Project Overview
Much can be learned from nature. This project explores the advantages of designing robots by mimicking life (biomimetics). Particularly, a class of robots have been created that mimick the design and control of the cockroach. Affectionately named "Sprawlita" after the sprawled leg configuration of many arthropods, the robots have been designed to show robust performance in unstructured environments with no or very little sensory feedback. The cockroach was chosen due to its well studied locomotion dynamics and robust design and performance characteristics. The most interesting feature of the robot is a passive rubber spring joint connecting the legs to the body. This joint, mimicking the springy, resilin lined joints of the insect, aids in disturbance rejection, accomplished without sensory feedback.


An important aspect of this project is the utilization of a new manufacturing process called Shape Deposition Manufacturing (SDM). The process allows parts to be embedded within a stiff polymer shell, allowing the robot to be more resistant to impact and disassembly. The process also allows dissimilar polymers to be molded together, making it possible to embed the soft polymer springs in the stiff polymer legs. The images on the left show a few steps of this process in the creation of the body of Sprawlita.

Our Role
Six different labs at four different universities are involved in this project (for more information visit the project homepage). At Harvard, we are focusing on designing a gripper that can be used in conjunction with the robot. As a departure from most robotic grippers, ours will incorporate active compliance through mechanical design rather than control strategies.

Similar to the robot, the gripper is inspired by arthropod grippers (or claws), which show a resemblance to traditional industrial robot end effectors (see right). However, arthropods have variable compliance in their joints due to the muscle articulation.
We have created a simple, two-fingered gripper in order to test the effect of changing the joint stiffness (varying compliance). This is accomplished by changing a spring that is in parallel with the joint, which connects an articulated finger to a passive finger. Initial results have shown that the gripper with compliant fingers has a significantly larger grasping space than that of the traditional stiff gripper (see technical reports below).

Here are two videos demonstrating properties of the simple gripper prototype: stiff vs. compliant fingers and compliant disturbance rejection


Future Work
The thrust of the future work on this project will be the implementation and evaluation of compliance in robotic grasping and manipulation. We will experiment with concepts such as variable stiffness and variable center of compliance, seeking to find a mechanical design solution to the problem of grasping in unstructured environments. We are identifying the task domain which will be used to design an appropriate gripper for use in conjunction with Sprawlita. Basic sensing technology will be implemented as well, providing position information, as well as indicating when contact with an object has been made. Eventually, these gripper designs will be produced using the SDM technology described above, and incoporated into the Sprawlita vehicle.

The ultimate goal is not to design a gripper to be used on a specific robot. We hope to show that compliance in grasping wins big when dealing with unstructured environments. Tasks such as undersea and space exploration, hazardous waste handling, and robotic surgery, in which a human operator is controlling a remotely located robot in an unfamiliar environment could be revolutionized by a technology that helps to make up for the lack of knowledge of the surroundings.

Technical Reports

Back to Past Projects