Team Vision for Detailed Design PhaseThe primary goal of this phase is to complete the detailed design of TigerBot 7, as well as, preliminary proof of concept prototypes to allow for easier implementation of concepts in MSD II.
- Design Goals
- FEA Validation of force-carrying parts
- Mechanical redesign for ease of manufacturability
- Completion of thigh drivetrain design (98%)
- Completion of hip support frame (96%)
- Completion of design of various incidental components (99%)
- Prototype Goals
- Manufacturing of Tekmonic plate for testing and verification
- Full or Subsystem-level FEA simulation
- Design Goals
- Completion of load cell integration design
- Completion of sensor network integration circuit design and PCB
- Completion of power distribution and sensing circuitry and PCB
- Completion of software topology design
- Completion of ROS topology
- Prototype Goals
- Preliminary prototype of motor control system.
- Prototype of ROS system
- Partial prototype/exploration of Gazebo/MoveIt simulation/integration
- Load cell integration design
- Responsible: Bryan and Collin
- Progress: Load cells have been obtained and preliminary brainstorming has begun. Electrical circuit prototype has been created with previous FSRs. Components need to be analysed for new load cells.
- Pending: Conditioning circuit design and validation method.
- Sensor Network Integration (Sensor to Teensy
- Responsible: Dan
- Progress: All communication system now known excluding radial encoder. Circuit design near complete pending the addition of load cell conditioning circuitry, radial encoder integration and mechanical stop switch integration.
- Pending: Connector selection and PCB design
- Power Distribution
- Responsible: Dan and Bryan
- Progress: Recessing power distribution topology to switch from daisy chained topology to distributed topology. This reduces the complexity of component selection for current sensing and would centralize current sensing components, reducing output wire run length from IC to Teensy. Preliminary component has been selected.
- Pending: Exploration of proper layout of a power PCB.
- Completion of Software Topology Design
- Responsible: Bryan, Dan and Felisa
- Progress: Preliminary topology has been theorized. This topology will change as development continues.
- Pending: Integration methodology of encoders and limit switches.
- Completion of ROS Topology
- Responsible: Felisa and Bryan
- Progress: Preliminary topology completed. Additional prototyping needed for development of message types and publication rates.
- Pending: Additional exploration into Gazebo/MoveIt implementation and methodology required. To be completed by Bryan over intercession.
- Prototype of Servo Control/Feedback System
- Responsible: All Members of EE and Required Support from ME
- Progress: Control system has been theorized. In process of ordering encoders and creating ROS topic/node system. ME team in process of creating simple test system.
- Pending: Creation of physical testing mechanisms, completion of control system code.
- FEA Validation of force-carrying components
- Responsible: John and Gus
- Progress: Ongoing
- Pending: Ongoing progess in simulation studies to optimize design for inteneded loads
- Redesign meeting for ease of manufacturability
- Responsible: ME Team
- Progress: Scheduled
- Pending: Feasibility of design simplifications
- Design of hip frame
- Responsible: ME Team
- Progress: Several concepts generated and modeled
- Pending: Ideation on general form of torso, analysis of expected stresses on hip plates
Power for Tigerbot is drawn from the main.The main supplies both TEKNIC power supplies, the Odroid, and both USB hubs. There is a power board on each limb consisting of three INA195 current shunt monitors. There is a 75V to 5V LM5017 buck converter. This is shown in the schematic below.
The INA195 current shunt monitors are placed in series with the power line that supplies each motor. There are three sensors in each limb. The INA195 can handle -16V to 80V. The output signal is fed to a Teensy for the feedback loop.
75v to 5v Buck Converter
A LM5017 buck converter is on each power board. The board takes in 75V and converts it to 5V to supply the encoders and transceivers. The board is still being configured for the correct settings. The buck converter is shown below under schematics.
Clearpath Integration Design
The clearpath motors require four control signals from the Teensy: A, B, Motor enable, and PWM. A and B dictate the direction of motion. Motor enable acts as an on/off switch for motor movement. The PWM signal controls the rate at which the motor turns based on its set frequency. All signals for clearpath motors are supplied by the Teensy and shifted through a transceiver.
Encoder Integration Design
The encoders are supplied by the 5V rail from the buck converter. The encoder takes in four data signals, SCLK, SSel, MISO, and MOSI. SCLK, MISO, and MOSI are all attached to each encoder and the Teensy. Each encoder has its own SSel that allows the Teensy to select which encoder to communicate with.
Force Load Cell Design
The four force load cells in each foot will transmit output data through a differential amplifier. The differential amplifier shown below shift the output voltage from 2.87V - 2.98V to 0V - 3.3V. The shifted output is fed into four Teensy analog ports.
ROS topics and code concerning the IMUs were developed using two libraries: i2c_imu and RTIMULib2-Teensy. Something to note is that since the IMUs used are all the same (the Adafruit 9DOF IMUs), they will have the same I2C addresses. The Teensy has two i2c busses available however, so two IMUs per Teensy will keep this from being a problem. If more IMUs per Teensy later becomes required, an i2c multiplexer could be used to give them different addresses.
Mechanical DesignsOver the course of the semester, countless hours have gone into the design and iteration of a humanoid robot. This process has allowed the robot to grow in complexity in necessary directions to really allow for the intended uses be possible. As more time was spent working with the designed model, compared to other industry examples, there were elements that needed to be heavily iterated to generate a complete robot assembly that will allow its intended use.
This humanoid robot has been designed with the optimization of humanoid characteristics and cost in mind. This has allowed for the engineering requirements drive by our customer to be met, while maintaining a budget that was agreed upon.
This design focuses on the assembly of six assemblies (hips, thigh, knee, shank, ankle, foot) into a complete robot assembly. Each subassembly is a complex assortment of off the shelf components and manufactured parts allowing for the motion freedom needed for the tasks this robot will be placed in.
This robot was built with as many off the shelf purchased parts as possible, as this allows the design to have completed parts used which allows more time and attention paid to the overall system level design and issue mitigation. These components include ball bearing, fasteners, ball screws, linear slides, and universal joints.
Bill of Material (BOM)
Force Load Cells
Tools Needed: Force load cell, DC Supply, DC voltmeter, varying weights up to 130 lbs.
Install the FLC into a stable rig to measure voltage change. Verify that there is a change in voltage when force is applied to the FSR. Measure various weights. Weights tested - 0, 10, 30, 60, 80 110, 130.
Run write operation to I2C Device and capture both SDA and SCL on oscilloscope. Compare clock to data transmitted. Measure the first 8 bits transmitted to 9DOF in comparison to SCL and record correct data transfer. Measure the returned 8 bits transmitted to the Teensy and compare to SCL.
Functional Test - Run data transfer to Teensy 200 times. Record number of times packets lost/data returned. If data transfer between two devices works, test rotation along the x, y, and z axes. Record actual to expected output for 1 degree increments. The 9DOF I2C device has been tested for both packet loss and repeatability. No additional support circuitry required.
Tools Needed: Teensy, SPI Encoder, Oscilloscope.
Connect SPI encoder to the 4 pin setup (MISO, MOSI, CLK, Enable) to the Teensy. Send packet to SPI encoder asking for current position. Record both CLK and data transferred. Compare timing of data packets to the change of the clock. Compare data returned to clock. Check the alignment of timing to clock to data returned. If both sent and received data align with clock, measure the output of the encoder in the serial monitor. Test changing the angle of the encoder for precision of the output. Set a new home position for the encoder and repeat the precision testing. Record results and measure sensitivity.
No additional support security required.
Teensy ROS Communication Limitation Assessment
Goal: Determine the limits of speed of communication between the Teensy and ROS, and what speeds would then be ideal.
-Port baud rate
-Message publishing rate
-Publisher and subscriber queue size
Success Criterion/Criteria: Teensy and ROS sends a heavy load between each other at a fast enough rate with no messages dropped and little latency.
-Have the Teensy subscribing and publishing data similar to what it’ll be expected to do, like data from IMUs, current sensors, etc. ROS would then take that data and do complex math with it, while checking if messages have been dropped or if there’s significant latency.
-Adjust baud rate and other variables until message dropping and such appears, then adjust variables to next-fastest testing, and run communication for a significant length of time.
OverviewThe intention is for validation of subsystem operation at the beginning of MSD II. Through subsequent testing of different subsystems; assembling the entirety of the robot should be straightforward and allow for idealized complete system testing. With the engineering requirements created from our customer revolving around a series of tests for complete robot design items, it is improbable and unlikely that testing can be conducted on the individual mechanical subsystems that would appropriately replace the complete system testing requirements.
Design and FlowchartsWith the continued design of the entire robot system, each subsystem (read "limb") is an element of the entire design. As seen by our images of the entire robot assembly, it does not add value to show each independent system for a testing plan.
- Updated assessment from Preliminary Detailed Design or link to other location. Have you driven the likelihood and/or severity down as you worked through the details of your design?
- Include a snapshot of your current risk assessment as well as a link to the live document.
MSD I Gate Review WorksheetLink to document on Google drive
Plans for next phase
- Finalize design of all circuits
- Begin design of PCBs
- Test data limits of Odroid and Teensy communication
- Work on simulation in Gazebo
- Continue Teensy library developing
- Test sensors with test rig and develop control structure using them
- Test and finalize PCB designs
- Use simulation to develop gait and path plans
- Complete developing ROS system and Teensy code
- Wire up electrical system on completed robot
- Extensively test finished robot and its electrical and software systems
- Complete drawing package for all parts that need to be fabricated
- Order all off the shelf components and raw material
- Major fabrication of components of each sub-assembly to allow for partial assembly during the intersession
- Iterated and completed loading simulations on critical components
- Outsourced machining in-process for 5-axis needed parts (ankle)
- Continue/complete fabrication of manufactured parts
- Modify manufactured parts for design changes that will be observed during assembly
- Assemble subsystems as soon as possible to hand off to the EE team for testing
- Assemble subsystems into full designed robot and work with EE team for gait testing