1. Potential Applications
Sensor Networks
Environmental
Earthquakes
Spill monitoring, pollution disasters
Wildlife monitoring
Disaster Response
Refugee Flow Monitoring
Public Health
Drug Delivery
Remote Disease Laboratories
Disposable medical tools

2. Challenges Dynamically evolve and generate new designs based on needs
Printing Actuators and Sensors
Fully Automating Fabrication
Reliability, Precision, and Durability
Disposability for potentially millions of devices

3. Collaboration + Planning
How can heterogeneous robots plan as a team?
Separate robots can reach states they couldn’t individually.
RRT variant? How do we incorporate cooperation into planning space?
Hybrid system trajectory planning (proximity -> modes)

4. Challenged in moving to a Multi-Robot Swarm
PPR Homework
John Romanishin
Challenged in moving to a Multi-Robot Swarm
Assembly time and cost:
Currently robots take many hours of tedious human assembly
Robustness:
Need to develop much more structures that endure many collisions and deployments and maintain reliabilitiy
Look into design and materials selection (more use of soft actuators/bodies?)
Power Sources:
Swarms of robots cannot realistically be individually charged/ cared for by people.
Look into environmental charging methods
Solar?
Inductive charging?

5. Control Design Fabrication Adaptability Capability
Robots can choose / create tools dynamically
What tools / materials to provide?
Capability
We can probably custom-design better tools, but requires prior task knowledge vs
Design of each robot may affect the designs of each other robot – can we automate this / let it evolve?
Iterative design with simulation / dynamical calculations?
Example: flying robot able to carry ground robot
Fabrication
Speed / Accessibility
Smaller, lighter robots often faster to create with few tools
Can we make them connect together to create something stronger?
Capability / Robustness
Larger, stronger robots often less disposable or more involved to create vs
Control
Autonomy / Emergent Behavior
Robots need to understand each other’s abilities
How to communicate this
Automatically dynamically assign tasks?
Robot capability may change as it reconfigures
New role in swarm? New tasks?
Global Knowledge / User Interaction
Central controller (maybe user) might be more informed but creates a bottleneck
Maybe tele-operated at first, but even this benefits from robot-specific interface
Auto-generate intuitive control of heterogeneous collective? vs
Joseph DelPreto /06/2015

6. Some ideas for PPR meeting by Shuguang Li 04/06/2015@CSAIL
Challenges of my current research
Print/fabricate soft robots with highly dynamic behaviors
Print out robots with programmable controllers
Robotics for good: Education, healthcare and social services.
1. Education: Origami robots for kids
2. Healthcare: Origami robots for stroke patients and Alzheimer's disease patients (Brain-hand exercises with fun robots)

7. REACT: Challenges Group-level programming abstraction
What are the right abstractions for group action?
Fault-tolerant coordination
Scaling verification to group action
Reasoning about time (logical vs real-time)
Reasoning about physical world (equations)

8. 3-D printable miniature factory for cooperative manipulation
Objectives:
1) to make miniaturized robotic manipulators using additive manufacturing systems
2) to demonstrate their independent motions for cooperation as in a factory system
Pneumatic actuation system
Micro-controller
sensors
Passive pneumatic actuator
My second project is a 3-D printable miniature-factory for cooperative manipulation.
The first objective of this project is to make miniaturized robotic manipulators using additive manufacturing systems…
The second one is to demonstrate their independent motions for cooperation as in a factory system. (~6m 50s)
Multi-material 3-D printed part.
Manufacturing system

9. Thoughts on contributing to a Swarm of Robots idea
The printed sensors may be useful for heterogeneous customized robots for better controlled activities/ interactions.
Pouch motors (as well as the printed sensor) are customizable, and materials used are accessible for general public. They can be integrated into simple ground robots.

10. Other ideas for Robots for Good
I believe the most advantageous feature of the robots from PPM is that they are largely customizable and easy to make by printing. If the design parameters can be defined directly and autonomously by the environment or the use case, the design of robots can be automatically generated and robots been automatically fabricated. To do this, I believe it would be worthy to exploit some concepts from machine learning, computer vision, etc.
One “extreme” example of the above thought: When some road cameras find that there is a crowd forming at a street crossing but no more details are provided, a robot (design is aerial if the street is full of people, legged if the robot needs to pass some staircase, or otherwise wheeled) carrying a camera folds up autonomously and approaches the crowd try to provide more information to the police.

11. 3-D printable miniature factory for cooperative manipulation
Objectives:
1) to make miniaturized robotic manipulators using additive manufacturing systems
2) to demonstrate their independent motions for cooperation as in a factory system
Pneumatic actuation system
Micro-controller
sensors
Passive pneumatic actuator
My second project is a 3-D printable miniature-factory for cooperative manipulation.
The first objective of this project is to make miniaturized robotic manipulators using additive manufacturing systems…
The second one is to demonstrate their independent motions for cooperation as in a factory system. (~6m 50s)
Multi-material 3-D printed part.
Manufacturing system

12. Thoughts on contributing to a Swarm of Robots idea
The printed sensors may be useful for heterogeneous customized robots for better controlled activities/ interactions.
Pouch motors (as well as the printed sensor) are customizable, and materials used are accessible for general public. They can be integrated into simple ground robots.

13. Other ideas for Robots for Good
I believe the most advantageous feature of the robots from PPM is that they are largely customizable and easy to make by printing. If the design parameters can be defined directly and autonomously by the environment or the use case, the design of robots can be automatically generated and robots been automatically fabricated. To do this, I believe it would be worthy to exploit some concepts from machine learning, computer vision, etc.
One “extreme” example of the above thought: When some road cameras find that there is a crowd forming at a street crossing but no more details are provided, a robot (design is aerial if the street is full of people, legged if the robot needs to pass some staircase, or otherwise wheeled) carrying a camera folds up autonomously and approaches the crowd try to provide more information to the police.

14. Mark Yim’s ideas for PPR projects
Single Robot Challenge (modified from Vijay’s slide)
Flying insect – a < 5 gram robot that can fly
(with limited autonomy, Matt Piccoli’s work)
Modular Robot Challenge (from Vijay’s slide)
Modules that can be assembled to form ground, aerial, or marine robots from specifications (Mark Yim’s sealpack)
Ground, aerial, and marine robots that can dock to build structures or macro robots
(e.g., Mark’s DARPA Temp project for the water, Raff’s arrays for the air)
Deep Thoughts about Challenges
Drag and drop interface to building complex robot systems that incorporate Mechanical Electrical and Task composition.

15. Challenges (Mickey) Challenges:
1. Incorporating sensors on robots in order to detect range and/or bearing information; necessary for formation control.
2. Collaborating and communicating between ground and aerial robots; knowing when hybrid robots are grounded or airborne for group hybrid system formation control.
3. Implementation from simulation to hardware.
Next Iteration:
Have a simulation of goal assignment, trajectory planning, and formation control problems wrapped together.

16. Vijay Kumar’s ideas for PPR projects
Single Robot Challenge
Fully autonomous, palm-size, < 50 gram flying robot
(autonomy includes state estimation, control, mapping, planning, interaction with remote human)
Flying insect – a 25 gram robot that can walk and fly
(with limited autonomy, recent collaboration between Rob and Vijay)
Swarm Challenge
Swarms of heterogeneous robots that can be tasked to (a) search, detect, localize target (e.g., IEDs); (b) construct a 3-D map within a cluttered environment of variables like temperature, hydrocarbon concentration
(ideas from iSmart proposal)
Modular Robot Challenge
Modules that can be assembled to form ground, aerial, or marine robots from specifications (Mark Yim’s sealpack)
Ground, aerial, and marine robots that can dock to build structures or macro robots
(e.g., Mark’s DARPA Temp project for the water, Raff’s arrays for the air)