Archive for category: Bipedal Robots

Christopher Andelin (Project Manager)

Mario Ramirez (Systems Engineer)

Qui Du (Manufacturing Engineer)

Andrew Laqui (Electronics and Controls Engineer)

Henry Ruff (Electronics and Controls Engineer)

Executive Summary

Program Objectives & Mission Profile

By Christopher Andelin (Project Manager)

RoFi is the fifth prototype from Project Biped.  It is a self-contained, bipedal robot that uses accelerometer feedback to balance. It has 12 DOF (degrees of freedom) and can walk around while avoiding obstacles using an ultrasonic range sensor. A small Android tablet in RoFi’s head provides the brains and an Arduino Mega provides the hardware interface.

The Design

The figures below are our mechanical design and layout of RoFi.

 

 

Project Features

Some key features of our design include:

• Redesigned head
• Battery Backpack
• Larger batteries for longer run time
• SMT PCB / Microcontroller
• Periscope provides vision
• RoFi must walk a figure 8 obstacle course

Some innovation of our design include:

• Mouse pad for shoes
• Designer glasses box for prototype head due to its ability to be shaped and retain form
• Servo locks are used to secure loose servos

 

System Design

System Block Diagram

By Mario Ramirez (Systems Engineer)

The Block diagram shows our system broken into sub-systems and how each sub-system is connected to complete the entire system.  Pins for the Atmega chip can be verified through our interface definitions.

 

Experimental Results

Accelerometer / Gyroscope

By Andrew Laqui and Henry Ruff (Electronics and Controls Engineers)

The accelerometer/gyroscope MPU-6050 found here was tested by implementing the test code found here and reading the raw values that were printed through the Arduino IDE. Detailed instructions on how to use those raw values can be found here in Henry Ruff’s blog post.

 

Ultrasonic Sensor

The ultrasonic sensor SEN136B5B was used on RoFi in order to detect and avoid objects. Various tests were performed in order to determine the range for which the sensor can detect. An in-depth explanation of these tests can be found here. The table below details the various angles that the sensor detected an object.

 

Servo Driver

The servo driver PCA9685 found here was tested because originally, an Arduino UNO was going to be used in order to reduce the size of the Arduino. Ultimately, rather than using an Arduino UNO, an ATmega1280 was implemented on the custom PCB design in order to accomplish this size reduction. A detailed description of how to servo driver was tested can be found here.

 

Experimental Results

By Mario Ramirez (Systems Engineer)

Below is our table of experiments completed.

 

Torque Testing

An analysis of the maximum torque needed to move each mass is done to insure RoFi’s servos can provide enough torque before stalling. Further information can be found here, https://www.arxterra.com/spring-2016-rofi-torque-report/.

This table shows the torque needed for each mass compared to the stalling torque of each servo.

 

Center of Mass

Center of Mass simulation was done to visualize RoFi’s center of mass with the new head design.  The center of Mass is represented by the black and white circle. Further information can be found here, https://www.arxterra.com/spring-2016-rofi-center-of-mass-report/.

Subsystem Design

Interface Definition

By Mario Ramirez (Systems Engineer), Andrew Laqui  and Henry Ruff (Electronics and Controls Engineers)

Below is our pin mapping for the ATmega 1280 chip.

 

Custom PCB / Microcontroller Design

Custom PCB Design

By Andrew Laqui and Henry Ruff (Electronics and Controls Engineers)

The beginning of the custom PCB design began with taking a look at how messy the physical breadboard was. Due to the number of servos (12), the wires needed to be carefully attached and handled in order to not have any wires disconnect.

Using the free software Fritzing, a digital diagram of the breadboard was constructed. Various libraries were added to the original software because the free software did not provide a few of the required parts. Those libraries can be found in the links to the Custom PCB Blog below.

Next, using the free software EAGLE, a schematic was constructed including all of the components that will be used on RoFi according to the interface definitions. The schematic was then used to determine the finalized layout for the actual PCB. More information for the PCB Layout can be found in Qui Du’s PCB Layout blog post.

 

Active Balancing of RoFi During Movement

By Henry Ruff (Electronics and Controls Engineer)

A large challenge for RoFi was the utilization of the MPU-6050 Accelerometer/Gyroscope while it was performing its active walking cycle. Because RoFi had to traverse a threshold, incline, and perform a figure-8 on such, it became very time-consuming and difficult to successfully replicate in separate run-throughs. As such, utilization of the MPU-6050 would have been able to make RoFi’s walking more reliable and easier, however, we were unable to successfully implement in. Instead, the following blog post elaborates on relevant efforts.

 

PCB Design

By Qui Du (Manufacturing Engineer)

I created the PCB layout once I received the PCB schematic from the Electronics and Controls Engineers. More detail about our PCB Design could be found here.

 

Soldering the PCB

After generating Gerber files and ordering the PCB components, I sent the Gerber files to class president Ryland Watts. One and half weeks later, we received the PCB board and PCB components. I soldered the PCB board with help from the class president Ryland Watts and the manufacturing division manager Juan Mendez. Here is the result of PCB board before and after soldering.

 

Hardware Design

My design decreases the size of the head and feet. Below is the completed SolidWorks 3D modeling of RoFi.

Detailed information about the design could be found here

After generate the STL files and making STL repair using netfabb Basic, all parts were ready to print. Here is the final result of RoFi.

 

Software Design

Software Block Diagram

By Mario Ramirez (Systems Engineer)

Here is a flow chart of our software.  The grey boxes are scripts written to handle the telemetry.  This code is given to you by Hill.  Blue squares represent the different subroutines for our code.  The red boxes represent the type of pin each sensor or actuator is connected to.  The orange squares represent the name and number of the pin each sensor or actuator is connected to.  The yellow boxes represent our sensors and actuators.  Green boxes represent what the user is controlling based on the software.

 

Verification & Validation Test Plan

Verification and Validation are based off requirements that can be found here, https://www.arxterra.com/spring-2016-rofi-preliminary-design-documentation/

Note: The Pass / Fail and Percentage Completed columns is our teams opinion and not that of the customer.

 

Project Update

Work Breakdown Structure

By Christopher Andelin (Project Manager)

Below is our work breakdown structure.

 

Resource Reports

Mass Report

By Mario Ramirez (Systems Engineer)

Project allocation is based off of the maximum mass the servos would be capable of moving before stalling.

 

Power Report

Project allocation is based off of a 15 minute run time.  This report assists in calculating what ratings you need for the batteries.

 

Budget Report

By Christopher Andelin (Project Manager)

Note: Our project cost does not include the PCB / Microcontroller and the components because our President bought them.

 

Schedule

Below is our project schedule.

View post on imgur.com

Schedule Summary

Here is my assessment on our overall progress:

• Mechanical Design 100% 
• PCB / Microcontroller Design 100% 
• Repaired issues with loose frame 90%
• Improved running time 100%
• Periscope lens 100%
• Polyfuse SMT 100%
• Organize cables 100%
• Moving parts hazard 0%
• Walking in a circle on flat surface 100%
• Surpass leveled surface to inclined surface threshold 100%
• Walking in a circle on inclined surface 25%
• Figure 8 obstacle course 30% (leveled surface circle only)

My overall assessment is that we are about 80% complete.

 

Burndown Chart

Our burndown chart dips in the last week due to getting the printed parts and PCB / Microcontroller implemented into our design.  We still need to complete the figure 8, hence the reason why we are behind schedule.

 

Concluding Thoughts

Since it is the end of the semester, I have some suggestions for future RoFi teams:

• Get approval for all purchases – Due to time constraints, our team was in a rush to get parts in to begin implementation.  We focused our attention on completing our mission objective without considering our customers expectations.  It is important to complete the objective as well as pleasing the customer.
• Replace servos with better quality – We had servo issues throughout the semester. We always had a spare servo in hand due to the unpredictability of servo failure.
• Make RoFi lighter – The servos would spasm causing RoFi to fall.
• Lower the center of mass – Due to RoFi being so top heavy, it was difficult implementing walking on an inclined surface.
• Avoid using Robot Poser – Robot Poser only worked on one of our team members laptops and we never found out why.  Robot Poser was also difficult to get it to start.  You may have to open task manager at times and close programs, plug and unplug the Arduino cable and the battery; I felt like we had to jump through hoops to get it to work.
• Implement the ultrasonic sensor, accelerometer/gyroscope and cordless walking – Due to customer request, we focused a lot of our attention getting RoFi to walk the figure 8 obstacle course and neglected other features.
• Get RoFi asap – We had to do a lot of documentation throughout the semester, we did not get to work on RoFi until about one month into the semester.
• Choose team members wisely – I was fortunate to have a great team but keep in mind that students have other classes and work to attend to.  Interview students and select members that have time, transportation and money.
• Project Manager – As Project Manager I had a lot of paperwork in the beginning of the semester. Work slowed down in the third month, so I dedicated a lot of my time getting RoFi to walk and assisted the engineers where I could.
• Be cautious of free 3D printing – Verify that the quality of the 3D printing material is to your liking.
• Understanding the customer – Realize that the customer has multiple projects and classes and will claim things were said or were not said that you may need to address.  Communicate and verify all concerns with the customer, student teacher and president.
• Update Mission Objective – Our mission objective said that the incline was 8 and 6 degrees;  we measured the incline and it varied from 14 to 7 degrees.  The room was also a burden to work in because the room was popular for labs and lectures.

 

Resource

Project Video

Bonus Videos

www.youtube.com/watch?v=96UEVOszRBs

Files

The following Google Drive contains useful resources for future RoFi teams, https://drive.google.com/drive/folders/0B_v74PMhdCbKSXVPQ285ekZEX28.

Christopher Andelin (Project Manager)

Mario Ramirez (Systems Engineer)

Qui Du (Manufacturing Engineer)

Andrew Laqui (Electronics and Controls Engineer)

Henry Ruff (Electronics and Controls Engineer)

Final RoFi 3D Modeling

Qui Du (Manufacturing Engineer)

Introduction

In Mechanical Design Rev2, https://www.arxterra.com/spring-2016-rofi-mechanical-design-rev-2/ RoFi’s head contains both the PCB and Arduino Mega.

I redesign the head to contain the PCB because my team implemented the Arduino Mega chip onto the PCB board. I also, changed to a smaller size screw for the head from M3 to M2. Since the size of the head covers is 5mm thick, having smaller screw holes give the frame more strength and they make the head look cleaner. Weight is also important therefore, I made the head as light as possible by carving holes, curving parts, and made some parts thinner.

 

Additional frames updated

Some of the servos come loose because they slide into the frame and do not attach to the frame; to fix this problem, I designed servo locks to lock the loose servos into place.

 

Mass Report Update

My new mechanical design has a mass of 1651.51 g which is below the mass report allocation of 1745 g found here, https://www.arxterra.com/spring-2016-rofi-preliminary-project-plan/. We prefer lower mass because our current mass in conjunction with our low quality servos, RoFi tends to spasm which causes RoFi to fall over.

 

Completed Design

Note: The Arxterra Logo is just for decoration and will not be on the final product.

All SolidWorks 3D modeling parts and assembly can be found here, Google Dive Link

Disclaimer: All parts were created in SolidWorks 2016; older versions of SolidWorks might not open the files. I included the STL files to allow students with older version of SolidWorks to view or print parts.

Christopher Andelin (Project Manager)

Mario Ramirez (Systems Engineer)

Qui Du (Manufacturing Engineer)

Andrew Laqui (Electronics and Controls Engineer)

Henry Ruff (Electronics and Controls Engineer)

PCB Layout

Qui Du (Manufacturing Engineer)

 

Once I received the PCB schematic from the Electronics and Controls Engineers, I create the PCB layout.

First, I clicked on the Generate/Switch to board   button at the top left corner of the schematic window. Then I clicked Yes on the Warning window to confirm that we will create a PCB layout from the schematic. The below window will appear:

 

Adjust PCB Board Size

I designed the PCB board the same size as the Arduino Mega (101.6 X 53.33mm) to allow the user to interchange the PCB with the Arduino Mega in RoFi’s head. To adjust the PCB board size, first I turned grid on by clicking the Grid button → changing the Unit to mm → selecting On at Display → and then by clicking Ok. Then I clicked on the top right corner of the white rectangular and adjusted the until shape until it reached 101.6mm X 53.33mm.

 

Visualize and Understand PCB Design

Since our level one requirement requires that we not use the Arduino Mega; we installed the ATmega1280 chip onto our PCB board. We still need to adjust the pin mapping from the ATmega2560 board to the ATmega1280 chip on the system block diagram. More information can be found at, https://www.arduino.cc/en/Hacking/PinMapping2560.

 

Component placement

The power trace takes up a lot of space on our PCB, therefore, it is better to isolate the power supply with other components.  In my design, I placed the power trace on the top border of the PCB board.

I began placing the ATmega chip, Bluetooth, USB chip, Accelerometer/Gyroscope onto the board and placed all the supporting components around them. Then I used smash features to smash and delete unimportant names on the PCB layout.

 

Make a Ground Plane and Copper Pours

Apply polygon on top layer.

Next click the Ratsnest button to view the ground plane.

Apply Polygon in bottom layer.

Click Ratsnest button.

 

Draw a Wire Signal

Determine a current width trace

According to our power report found here, https://www.arxterra.com/spring-2016-rofi-preliminary-project-plan/ each servo can draw up to 1.896A.

Since we plan to order our PCB board from www.4pcb.com, I used the trace width calculator from their website to determine the trace width of each wire.

Note: The thickness of the trace cannot be adjusted in Eagle CAD. We can order the trace thickness when we send the PCB layout to the PCB fabrication house. In our PCB layout, we ordered a 2 oz/ft^2 thickness.

 

DRC Check

Each PCB fabrication house has its own set of design rules, since we are using www.4pcb.com to fabricate our PCB, we followed their design rules.

4pcb Fabrication House DRC Check

Step 1

Download the DRC file by going to,  http://colinkarpfinger.com/blog/2010/ordering-pcbs-designed-with-eagle/.

Right click typical_4pcb.dru → choose Save link as → Save the file

Step 2

The window below appears once the DRC check icon is selected in the board window.

 

Create Gerber Files

Step 1

Enable vector front by going to Option → User Interface → check Always vector front → OK

Open the CAM job → choose the CAM job file → click Open → click Process Job to generate Gerber files.

 

Get PCB Layout Quote From Fabrication House

After I zipped all the Gerber files and uploaded them to, www.4pcb.com, I received the PCB Layout Quote.

 

Christopher Andelin (Project Manager)

Mario Ramirez (Systems Engineer)

Qui Du (Manufacturing Engineer)

Andrew Laqui (Electronics and Controls Engineer)

Henry Ruff (Electronics and Controls Engineer)

Bluetooth Communication

Mario Ramirez (Systems Engineer)

To meet requirement 1.8, RoFi must communicate with the Arxrobot App and have an on and off state using the users phone.  Begin your Bluetooth connection by downloading the Arxrobot firmware, https://github.com/arxterra/BiPed-Robot.  This code has the commands and command decoder subroutine to control RoFi using the Arxrobot app.

First, connect and setup your Bluetooth mode, HC-06.  The pass code for connecting to the module with your phone is ‘0000’.

 

Within the Arxrobot firmware a custom command will be made.

To initially test the custom command, a simple LED on and off command will be implemented within the command decoder subroutine.

Once proper Bluetooth connection is tested, change the implemented command to your desired action.  For RoFi, we created a trigger after the command is received.

When the command packet is sent, the code will then read the 3^rd byte.  If the byte is HIGH it will set trigger to HIGH.  If the byte is LOW it will set trigger to LOW.  Using this trigger variable, an “if” statement was implemented into our main loop code.

Figure 6 shows two “if” statements for different values of the trigger.  When the trigger is HIGH, it will light an LED and put RoFi into an Update Action subroutine which allows RoFi to update to its next frame.  When trigger is set to LOW, RoFi will not take any action and will remain at its current frame until trigger is set back to HIGH.