Project :: ECE 445 - Senior Design Laboratory

Project

#	Title	Team Members	TA	Documents	Sponsor
38	NannyBot for Robots Developing to Walk	Alejandro Diaz De Argandona Araujo Chulwon Choi Karthikeyan Sundaram	Vassily Petrov	design_document1.pdf design_document2.pdf design_document3.pdf design_document4.pdf design_document5.pdf final_paper1.pdf photo1.jpg proposal1.pdf proposal2.pdf
	Problem: A common way for robots to learn to walk is through trial and error. The robot walks, falls, records the failure, and walks again and repeats the process until it learns to walk stably. However, this learning process requires many failures. This process is tedious, as people need to manually bring it up and back to the starting place. It can be costly if the robot becomes damaged when it falls, and even dangerous if a person makes a mistake when trying to put it back up and bring it back. There are methods that make the robot, such as installing a motor and strings to reel in the robot when it falls. However, these methods require installation, which may not be possible in all environments. Solution Overview: The process of robots learning to walk would be greatly simplified and much safer if it was aided by another robot, specifically designed for the purpose. Our idea is to create a robot, namely a NannyBot, that can aid the robots learning to walk. Instead of humans, our robot would follow the robot learning to walk, and pick the said robot up when it falls. It would be a wheeled robot, so that it won’t fall over, and it would have a strap tied to the walking robot, and lift the robot using the strap. This robot will be controlled by a human user by a wired controller (such as a video game controller like an Xbox controller), who will use the robot to follow the walking robot and bring it back when it falls. It would not require installation anywhere, and therefore be relatively unrestricted by the environment. The robot will have omni-directional wheels, allowing it to follow any direction the robot may go. The NannyBot will have a skeletal box shaped structure, with one side open, and have 4 strings attached to a strap attached to the walking robot. When the human controlling the NannyBot pushes a button, the 4 strings will wind, bringing the walking robot up into the air. The NannyBot will then return with the walking robot the where it started. The target of this Nannybots are walking robots of size approximately 30cm30cm50cm, and 5kg heavy. A possible robot to use this on is Robotis OP2, which Professor Kim Joohyung, the professor who pitched this project, has. The Robotis OP2 has a default walking speed of 24cm/sec, which we will aim to match. Originally, the idea was to have this done completely autonomously, but the TAs we talked to told us that the idea was beyond the scope of the class, and that we needed to simplify it. The structure of the NannyBot may be difficult to understand, so we attached a of the rough drawing of the structure of the Nannybot as photo. Solution Components: - - The NannyBot will have straps attached to it, used to lift the walking robot when it falls. The straps will be connected to the walking robot as well. There will be four strings, each on an upper corner of the NannyBot, which will bring up the walking robot stably when it is lifted. - The structure of the NannyBot will have a box shape, partially surrounding the walking robot, so that the NannyBot can withstand the walking robot’s weight as it lifts the walking robot - Aside from the straps, NannyBot’s dimensions will be big enough from the walking robot so it will not impede the walking robot’s path, or have the walking robot crash into the NannyBot when it falls. - We will use omni directional wheels for the NannyBot, so that it can follow the walking robot in any direction. - We will require powerful motors with encoders for lifting the walking robot, and powerful motors for the wheels as well, as the NannyBot is expected to return to the beginning place with the walking robot. Criterion for Success: - Provide a mechanical structure that can support the robot. - Have the NannyBot move in direction of robot with the help of a controller. - Have the NannyBot be unable to accidently crash into the walking robot. - Upon receiving a command from the controller, the NannyBot should be able to pick up the robot. - The NannyBot should be strong enough to bring the walking robot back to the starting point. Extended goals: - Have the NannyBot automatically track the robot. - Have the NannyBot sense when the robot has fallen down and pick the robot up after this detection.

Title

Team Members

Documents

Sponsor

NannyBot for Robots Developing to Walk

Alejandro Diaz De Argandona Araujo
Chulwon Choi
Karthikeyan Sundaram

Vassily Petrov

design_document1.pdf
design_document2.pdf
design_document3.pdf
design_document4.pdf
design_document5.pdf
final_paper1.pdf
photo1.jpg
proposal1.pdf
proposal2.pdf

Problem:
A common way for robots to learn to walk is through trial and error. The robot walks, falls, records the failure, and walks again and repeats the process until it learns to walk stably. However, this learning process requires many failures. This process is tedious, as people need to manually bring it up and back to the starting place. It can be costly if the robot becomes damaged when it falls, and even dangerous if a person makes a mistake when trying to put it back up and bring it back. There are methods that make the robot, such as installing a motor and strings to reel in the robot when it falls. However, these methods require installation, which may not be possible in all environments.

Solution Overview:
The process of robots learning to walk would be greatly simplified and much safer if it was aided by another robot, specifically designed for the purpose. Our idea is to create a robot, namely a NannyBot, that can aid the robots learning to walk. Instead of humans, our robot would follow the robot learning to walk, and pick the said robot up when it falls. It would be a wheeled robot, so that it won’t fall over, and it would have a strap tied to the walking robot, and lift the robot using the strap. This robot will be controlled by a human user by a wired controller (such as a video game controller like an Xbox controller), who will use the robot to follow the walking robot and bring it back when it falls. It would not require installation anywhere, and therefore be relatively unrestricted by the environment. The robot will have omni-directional wheels, allowing it to follow any direction the robot may go. The NannyBot will have a skeletal box shaped structure, with one side open, and have 4 strings attached to a strap attached to the walking robot. When the human controlling the NannyBot pushes a button, the 4 strings will wind, bringing the walking robot up into the air. The NannyBot will then return with the walking robot the where it started. The target of this Nannybots are walking robots of size approximately 30cm*30cm*50cm, and 5kg heavy. A possible robot to use this on is Robotis OP2, which Professor Kim Joohyung, the professor who pitched this project, has. The Robotis OP2 has a default walking speed of 24cm/sec, which we will aim to match.
Originally, the idea was to have this done completely autonomously, but the TAs we talked to told us that the idea was beyond the scope of the class, and that we needed to simplify it.
The structure of the NannyBot may be difficult to understand, so we attached a of the rough drawing of the structure of the Nannybot as photo.

Solution Components: -
- The NannyBot will have straps attached to it, used to lift the walking robot when it falls. The straps will be connected to the walking robot as well. There will be four strings, each on an upper corner of the NannyBot, which will bring up the walking robot stably when it is lifted.
- The structure of the NannyBot will have a box shape, partially surrounding the walking robot, so that the NannyBot can withstand the walking robot’s weight as it lifts the walking robot
- Aside from the straps, NannyBot’s dimensions will be big enough from the walking robot so it will not impede the walking robot’s path, or have the walking robot crash into the NannyBot when it falls.
- We will use omni directional wheels for the NannyBot, so that it can follow the walking robot in any direction.
- We will require powerful motors with encoders for lifting the walking robot, and powerful motors for the wheels as well, as the NannyBot is expected to return to the beginning place with the walking robot.

Criterion for Success:
- Provide a mechanical structure that can support the robot.
- Have the NannyBot move in direction of robot with the help of a controller.
- Have the NannyBot be unable to accidently crash into the walking robot.
- Upon receiving a command from the controller, the NannyBot should be able to pick up the robot.
- The NannyBot should be strong enough to bring the walking robot back to the starting point.

Extended goals:
- Have the NannyBot automatically track the robot.
- Have the NannyBot sense when the robot has fallen down and pick the robot up after this detection.

Featured Project

**Partners (seeking one additional partner)**: Braeden Smith (braeden2), Srivardhan Sajja (sajja3)

**Problem**: We imagine this product would have a primary use in military/law enforcement application -- especially in dangerous, high risk missions. During a house raid or other sensitive mission, maintaining a quiet profile and also having good situational awareness is essential. That mean's that normal two way radios can't work. And alternatives, like in-ear radios act as outside->in communication only and also reduce the ability to hear your surroundings.

**Solution**: We would provide a series of small pocketable devices with long battery that would use LoRa radios to provide a range of 1-5 miles. They would be rechargeable and have a single recessed soft-touch button that would allow someone to find it inside of pockets and tap it easily. The taps would be sent in real-time to all other devices, where they would be translated into silent but noticeable vibrations. (Every device can obviously TX/RX).

Essentially a team could use a set of predetermined signals or even morse code, to quickly and without loss of situational awareness communicate movements/instructions to others who are not within line-of-sight.

The following we would not consider part of the basic requirements for success, but additional goals if we are ahead of schedule:

We could also imagine a base-station which would allow someone using a computer to type simple text that would be sent out as morse code or other predetermined patterns. Additionally this base station would be able to record and monitor the traffic over the LoRa channels (including sender).

**Solutions Components**:

- **Charging and power systems**: the device would have a single USB-C/Microusb port that would connect to charging circuitry for the small Lithium-ion battery (150-500mAh). This USB port would also connect to the MCU. The subsystem would also be responsible to dropping the lion (3.7-4.2V to a stable 3.3V logic level). and providing power to the vibration motor.

- **RF Communications**: we would rely on externally produced RF transceivers that we would integrate into our PCB -- DLP-RFS1280, https://www.sparkfun.com/products/16871, https://www.adafruit.com/product/3073, .

-**Vibration**: We would have to research and source durable quiet, vibration motors that might even be adjustable in intensity

- **MCU**: We are likely to use the STM32 series of MCU's. We need it to communicate with the transceiver (probably SPI) and also control the vibration motor (by driving some transistor). The packets that we send would need to be encrypted (probably with AES). We would also need it to communicate to a host computer for programming via the same port.

- **Structural**: For this prototype, we'd imagine that a simple 3d printed case would be appropriate. We'd have to design something small and relatively ergonomic. We would have a single recessed location for the soft-touch button, that'd be easy to find by feel.

**Basic criterion for success:** We have at least two wireless devices that can reliably and quickly transfer button-presses to vibrations on the other device. It should operate at at *least* 1km LOS. It should be programmable + chargeable via USB. It should also be relatively compact in size and quiet to use.

**Additional Success Criterion:** we would have a separate, 3rd device that can stay permanently connected to a computer. It would provide some software that would be able to send and receive from the LoRa radio, especially ASCII -> morse code.

Project

Covert Communication Device

Featured Project