Soft robots have a continuously deformable structure with muscle-like actuation that emulates biological systems and results in a relatively large number of degrees of freedom as compared to their hard-bodied counterparts. Soft robots have the potential to exhibit unprecedented adaptation, sensitivity, and agility. Soft bodied robots promise to 1) Move with the ability to bend and twist with high curvatures and thus can be used in confined spaces; 2) Deform their bodies in a continuous way and thus achieve motions that emulate biology; 3) Adapt their shape to the environment employing compliant motion and thus manipulate un-modeled objects, or move on rough terrain and exhibit resilience; 4) Execute rapid, agile maneuvers, such as the escape maneuver in fish. We are developing new design, fabrication, modeling, control, and planning algorithms for soft robots.

Our new soft robots include a soft robot fish, a soft arm capable of manipulation in planar environments, and a soft dynamic arm capable of dynamic manipulation in three dimensional environments.

The objective is to develop data-driven customized transportation. We are developing a fleet of self-driving cars consisting of golf carts and electric cars. Mobility on demand aims to transform transportation into a utility, to be available to anybody, anytime. If you want to go somewhere, you book a ride, the robot car comes to where you are and drives you where you want to go. After dropping you off, the robot car coordinates with the other robots in the system to determine where its next pickup location is and drives itself there. An optimization engine ensures that people’s waiting times, car trajectories, and vehicles in the system are optimized.

In October 2014 we conducted a public trial at the Chinese Gardens in Singapore. We invited the public to test our mobility on demand system. During the week long trial over 500 people booked and took rides. The robot cars navigate successfully, following the road, avoiding pedestrians, bikers, and monitor lizards, and bringing passengers to their selected destinations.

Here is our Self-Driving Car Team:

and some videos from our work:

Rebalancing Cars

Autonomous Driving

Many different types of robots are available today, but each of these robots took many years to produce. The computation, mobility, and manipulation capabilities of robots are tightly coupled to the body of the robot—its hardware system. Since today’s robot bodies are fixed and difficult to extend, the capabilities of each robot are limited by its body. Fabricating new robots, add-on robotic modules, fixtures, or specialized tools to extend capabilities is not a real option, as the process of design, fabrication, assembly, and programming is long and cumbersome. We need the design and fabrication tools that will speed up the fabrication of robots. Imagine creating a robot compiler that takes as input the functional specification of the robot (for example “I want a robot to play chess with me”) and computes a design that meets the specification, a fabrication plan, and a custom programming environment for using the robot. Many tasks big and small could be automated by rapid design and fabrication of many different types of robots using such a robot compiler.

$ emacs myrobot.rbt

“I want a robot to play chess with me”

$ make myrobot

Parsing specification …done

Determining behaviors …done.

Generating mechanisms …done.

Assembling components …done.

Printing …done.

Success!

Our current solution to the robot compiler is a data-driven approach to design. Existing robot designs and supporting control algorithms that sit in a database are segmented and composed to create new robots. The user imagines a machine, say a duck robot or an ant robot and defines its behaviors (say a robot should pick up a piece and move it), 2) assemble mechanisms into an integrated design and 3) rapidly fabricates the device and its programming and control substrate.

Title of Video

The idea is to make a module or voxel that can move to sit on another voxel, then another, so that it can build different shapes. And eventually, the modules can turn themselves into a support beam, a tool, a robot with a different number of limbs, etc. We are developing new designs for self-reconfiguring robots and new planning and control algorithms to control the reconfiguration of the robot and its motion.

We have developed a several approaches to self-reconfiguring robots including The Molecule Robot, the Crystal Robot, the Shady Robot, the Pebbles, and M-blocks.

M-blocks are robot cubes.

M stands for motion, magnet, and magic. Motion because the cubes can move by jumping. Magnet because the cubes can connect to other cubes using magnets and once connected they can move together; the cubes can also connect to assemble structures. Magic because the surface of the cube is very clean, we do not see.

Title

Optimal Coverage With Quadrotors

Multirobot Navigation

Informative Path Single Robot

Robots have already enabled mass production and robots together with new computational systems have the potential to enable mass customization. Today, robots are used extensively in manufacturing. But manufacturing systems of today have very limited flexibility, often requiring months of fine-tuning before an industrial assembly line is ready for production. We envision the manufacturing systems of the future, in which agile, flexible teams of mobile robots coordinate to assemble complex and diverse structures autonomously. This approach has the potential to meet the demands of modern production: ever- shortening product life-cycles, customized production, and efficiency. We are developing algorithms and systems that allow multi-robot system to collaborate in complex assembly tasks that require multiple operations.

Asking for Help

We wish to design the body of the robot flat. We wish for the robot to have built-in intelligence, and ability to move in the world. Once designed, the robot is printed and a special control system is used to deliver a 3d functional prototype. In this project we explore how “baking” the robot by uniform heating can be used to self-fold the robot. The secret sauce is in the composition of the body as a three layer structure. The top and bottom layers are structural. The middle layer is the bending control layer—made form material that reacts to heat and shrinks. By cutting gaps in the top and bottom layers we can control the bending angle of the material. For example, if the gap of the top layer is wider, the hinge folds into a bigger fold angle. With this idea we can create a self-folding compiler. The input is the image of the object we’d like to create. Standard techniques are applied to generate a simplified mesh, which is unfolded. The unfolding revels the angles needed to create the 3d structure and these angles are mapped to gaps in the top and bottom layers of the material used to make the robot.

The result is files that can be sent directly to a laser cutter for fabrication. We can fabricate a multitude of robots that can self-fold and then walk, swim, manipulate objects, and self-recycle when the robot’s mission is completed.

We are developing algorithms and systems to extract the salient points in data, and map the signals into words and abstractions; for example GPS points can be mapped street addresses, and street addresses can be connected to activities such as “having coffee at Starbucks” using online repositories such as the Wikipedia and Yelp. To address the data summarization challenge we are developing a data summarization based on coresets. Coresets allow us to run slow algorithms on large data by carefully selecting a small subset of the data, which is guaranteed to give approximately the same result as the entire data set (but running the algorithm on the entire data set is intractable.) These coreset points summarize the large data clusters as well as outliers contained in small clusters. We have developed coreset algorithms that can summarize complex signals including video and GPS signals. Using these algorithms, we are developing a system called iDiary. In iDiary a system can provide their GPS or Video data streams. These data streams are summarized and mapped into textual descriptions of the trajectories. The iDiary system includes a user interface similar to Google Search that allows users to type text queries on activities (e.g., “Where did I buy books?”) and receive textual answers based on their GPS signals.

Title