Group 18-01 Ball Balancing Robot: (from left to right) George Risi, Dan Mariano, Kevin Simone, Jeffrey Vo, and Tyler Ney-Smith after the final presentation.

Poster made for the presentation after working on the project for 10 weeks.

This is the poster we created for the project at the end of the two weeks. 

Week 10 Update

This week the final assembly was completed. The bushings were screwed onto the motor shafts. The wheels were pressed onto the bushings. The microcontroller was mounted on the top plate via acrylic brackets. A better suited ball was chosen for a better frictional surface. 

New ball with the robot on top.

Microcontroller mounted on the top plate

 Side view of the Microcontroller mounting brackets

 Top view of the Microcontroller mounting brackets

Weekly update: Week 9

We completed a new motor mount without a crack. The bends look a lot more precise in this one than the last one.

Newest mount with motors

The group also lathed the bushings out of billet aluminum. Now they need to be drilled to the proper size on the inside and another hole drilled on the side and tapped to hold the set screw. The set screw will be used to hold the busing/wheel to the shaft.

Lathed bushings

Close up view

Weekly update: Week 8

Finished line bending the motor mount.

Bracket with the bottom plate

Bracket with both plates and motors

Robot resting on ball without wheels. With wheels on, the plexiglass will not contact the ball.

The group had some trouble with a corner cracking. Decided to start making a new one with the other plexiglass cut out we had.

This is a detail of the crack in the motor mount. It was caused by an uneven tightening of the screws that hold the motor to the mount.

Weekly Update: Week 7

Machined thicker, wider acrylic. thickness = .118 " and width = 39 mm

This week: Machined with thicker plastic, worked on bushing for motor to wheel contact, started hot wire bending, microcontroller research

Total machining time for the laser cutter ( for top platform, bottom platform, and 2 brackets) = about 30 minutes


Machining Process Documented:

Main screen of CorelDraw(laser cutting software) and Top Platform cutting

Cutting circles out of the main brackets

Started working on bending the material into shape with the hot wire bender:

Upper platforms assembled:

Assembled platforms next to bocce ball for comparison:

Bending process in action: 

Closer view of bent corners:

Electronics Circuit

The group looked into how to provide power to the circuit. They found out that the Arduino UNO controller they had been considering using draws 7-12v. The motors draw 2.7v and the uno only outputs 5v. To run all three motors each one will have to be run through an A4988 stepper motor driver carrier, available from pololu. This device can receive logic voltage commands from the arduino but also receives voltage from the power supply to power the motors. It uses between 8 and 35v. We need three of them, one for each motor. The motors then get their power from the batteries, using the motor driver as an intermediary. The motor driver receives a logic voltage from the micro controller that regulates how much voltage to run to the motors.

Week 6 Update

Week 6 update:
revised cad drawings
parts arrived
machine shop work



The cut acrylic: the holes are cut through, and the lines are etched for the line bender.

Bocce Ball next to the motor for size comparison

Motor closer up pictures (35mm x 35mm x 36 mm [depth]) M3 screw holes

Wheel side view.





















Wheel top view.





















M3x 6 screws ( slightly too long for our motors- ordering M3x5)

Holes are slightly misaligned due to an cad error. Model revised and holes now match. Etching (for bend lines) showed up well, and are correctly spaced.

2 more views of the plate against the motor.

View of the inside diameter of the wheels versus the motor shaft. Working on lathing aluminum to press fit between the shaft and wheel. 

Ball Contact: Coordinates

           As we begin the construction the general frame of the Ball Balancing Robot, it is essential that one gathers as much information as possible. An important piece to the construction is to find where the ball is in contact with the motors. In finding the point of contact we can begin sketches and construction around these three important coordinates.

          These coordinates could be easily found by using a series of algebraic moves. By knowing the angles that separate each motor in three dimensional space and the radius of the ball these coordinates can be solved for.

          The following work is shown below: 

New Motor (motor 2)

motor

Updated Motor Mount

After determining that our previous design for the motor mount would not be strong enough to withstand the total torque, we redesigned the mount as pictured below. The single-piece design mounts at three points on the robot's lower platform and connects in the middle. It will be made of bent acrylic that wraps around the motor and attaches using the four mounting screws surrounding the shaft.

The perfect stepper motor

http://www.pololu.com/catalog/product/1209

This hybrid bipolar stepping motor has a 1.8° step angle (200 steps/revolution). Each phase draws 500 mA at 10 V, allowing for a holding torque of 1000 g-cm (14 oz-in).

Overview

This hybrid bipolar stepping motor has a 1.8° step angle (200 steps/revolution). Each phase draws 1000 mA at 2.7 V, allowing for a holding torque of 1400 g-cm (20 oz-in). The motor has four color-coded wires terminated with bare leads. Our 5mm universal mounting hub can be used to mount objects on the stepper motor’s 5mm-diameter output shaft.

Specifications

• Size: 35 mm square x 36 mm, not including the shaft (NEMA 14)
• Weight: 180 g (6 oz)
• Shaft diameter: 5 mm
• Steps per revolution: 200
• Current rating: 1000 mA per coil
• Voltage rating: 2.7 V
• Resistance: 2.7 Ohm per coil
• Holding torque: 1400 g-cm (20 oz-in)
• Inductance: 4.3 mH per coil
• Lead length: 12 in

These specs and pictures are straight off of the Pololu website.  Our team thinks that this motor is perfect for our robots drive motors because of it's really high torque at low speeds.  We think that this is one of the most important features in selecting a motor for our robot because we are concerned about the robot tipping over more when it has to make small adjustments because momentum will not be available to help keep the robot stable.  

finding the right ball for the job

A very large component of our ball balancing robot is the the physical ball.  Below is our primary list of constraints for the robot's ball:

1. We need to select a ball that will be able to support the weight of the robot without contorting too much.
2. The ball needs to have enough traction that the slippage will not become a problem with the hi torque motors on smooth surfaces
3. The ball's surface needs to be smooth enough that the omni wheel does not skip on seams of the ball.

Basic Design

Below are screen shots of the basic mechanical design of the robot.

 Side View 

 Bottom View

Top View

Gyroscope and Accelerometers

For this robot to be able to balance the most important component to this task is not the body of the robot but the data that is drawn from the robot. It is like a body without a brain it serves no purpose with out the ability to function. These two parts to the robot will be able to extract information sending it though the micro chip and later processing it.

Accelerometer:
An accelerometer is a device that has the ability to measure and calculate accelerations in micro-movements. With in this device it is broken down to a simple structure that can detect three dimensions of acceleration. It is composed of springs holding together a centerpiece. When an acceleration is felt, a certain amount of voltage is generated though kinetic energy. The accelerometer has read and record the acceleration felt by the device.

Gyroscope:
A gyroscope is a device that has the ability to measure and calculate angular acceleration based on its orientation. The angular acceleration can be calculated due to the structure of the gyroscope. Similar to the accelerometer volts are read when the internal parts of the gyroscope are displaced. However the structure varies from a basic accelerometer. There are disk and wheels whose axis is free to take up any orientation. For example if a force is felt on the positive x axis then the disk will rotate clockwise. This is important because due to the object that this robot will be balancing on, there are endless possibilities of motion.

The issue of the Accelerometer is that it lacks the precision and data that a gyroscope has in addition the gyroscope has an issue with calibration. By combining these two devices one could eliminate the issue of accuracy in data.

Parts

Motor Dimensions

omni wheel dimensions

Motors

Stepper Motor Basics - wiki link

what motor is right for our application??

size constraints - what dimensions will work?

torque spec constraints- how much torque do we need?

Small omni wheels in the size we want

http://www.robotshop.com/dagu-omni-wheel-41mm.html

1.6 inch diameter

.551 inch hole in middle

$3.60 each

Other similar sizes available

Stepper Motors
http://www.robotshop.com/bipolar-stepper-motor-10v-0-5a.html

Chassis Materials Study

After stopping off in the machine shop our team learned that the most efficient material to use to build our robot would be Acrylic.  The machine shop’s capabilities to machine Acrylic are quite vast.  Acrylic bends when it is heated, to be shaped into any desired angle with minimal machining, and it is very inexpensive to purchase in extruded rods or sheets. 

                    -These are typical examples of sheet acrylic, the material is available in a variety of colors

                Acrylic’s physical properties are perfect for our application because it is very easily cut, it is more flexible than glass, it is more shock resistant than glass, it is abrasion resistant, and it is easily cleaned.  Our team will be able to utilize the ease of cutting that the acrylic offers.  We will be able to take advantage of the laser cutter at the Drexel machining shop as well as band saws and potentially hand saws.  We will also benefit greatly from the milling, drilling, threading, and tapping. 

                The team envisions the manufacturing process of the acrylic to encompass a large sheet of acrylic to be fed into the laser cutter, where it will be cut to precise dimensions.  The pieces then will undergo a series of drilling and tapping holes for screws to assemble the bot into a 3d shape.  The project is limited by the capabilities of the laser cutter and are restricted to ¼ inch acrylic at the thickest.  The team is currently researching the most efficient and precise ways to cut out the acrylic components for the chassis of the robot as well as the most effective use of either screws and/or epoxies to secure the components together. 

     This is an example of acrylic that has been cut with a laser cutter.  The edges of the acrylic are smooth and precise, this could potentially be the best way to create a chassis with our specific design constraints.    

A Curious Fabrication Technique

                    -http://lcamtuf.coredump.cx/rstory/

     This image was pulled off of a blog by a polish engineer that fabricates robot parts using a cnc mill to cut precise molds to cast pieces in.  The link above leads to his blog, which introduces his process in design, and is a very interesting read.  He does a great job evaluating the different materials that he has available and the evolution of his projects can be directly related back to our project.  

Designs and Information

Basic Designs:

From IEEE

The wheels are powered with NIDEC motors and micro-step controllers to achieve a 0.225 degree per step which resulted in smooth rotation.

The control system was run on a 16-bit microcontroller which received data from gyroscopes and accelerometers.

Robot Close up views:

note: dampeners

shocks and springs

ball rack