Team A1: Cooperative vs Non-cooperative Autonomous Driving – Page 3 – Carnegie Mellon ECE Capstone, Spring 2020: Kylee Santos, Tito Anammah, Serris Lew

Built Cooperative path planning environment class
- controls vehicle movements based on othetor vehicle locations and cooperative strategy
- Current fairness strategy prefers vehicles from a longer queue but can scale to a more complex strategy given a set of features.
- Environment continuously selects a single vehicle to pass through the intersection (of a figure 8 track)
- Path planning class allows for a different number of cars on either track
Integrated path planning class with graphic simulation so that at every time step it takes as an input the vehicles on the track and updates their locations.
Tested performance of graphic simulation with PP class and it seems to work fine
Fixed a bug where certain cars could pass through other cars
TODO
- Incorporate vehicle acceleration/deceleration to allow for smoother movement as opposed to the vehicles coming to an abrupt stop.
- Investigate better fairness strategies to decide vehicle to pass through the intersection

Decided on reasonable real world parameters based on papers read
Implemented the Intelligent Driver Model using these parameters
Analyzed the graphics and created a scale to convert pixels to meters
Cleaned up, modularized, and corrected graphics and vehicle code in existing codebase
Integrated the planning code with the graphics and vehicle code and tested
Worked on Risk Management Plan and updated Gantt Chart

Modularized code more to allow cars to turn right or left depending on which track its on
Personalized it so user sets cars initial location by clicking anywhere inside the track
- Automatically sets the angle from location on track
- Allows many different cases to test with no limit on number of cars
Display each cars data on the bottom of the screen
Working on getting a timer to find the relationship between vehicle class speed and rotations
- Will help in later cases to calculate throughput over a given period
Created new gantt chart
TODO:
- Make sure vehicles can’t overlap when adding to the track
- Create new tracks
- Develop vehicle class to make it more compatible with PP

Planned for added complexities for our repurposed project
Wrote the Statement of Work
Participated in 3 meetings with the group to discuss project
Helped design the graphics code (ie. modularize, integrate with path planning, etc.)
Continued to research non-cooperative path planning
- Ie. Trying to relate the purely simulated data such as pixels and angles with the equations that use real world measurements like meters per second.
TODO next week:
- Start to implement simple planning code that stops when obstacle is sensed.

Finalized new design to be simulated completely on software
- Revised each member’s tasks/responsibilities
- Discussed new goals and priorities
Completed Statement of Work
Conducted 3 meetings to devise a plan, split up responsibilities, and get some work started
Started designing graphics for simulation
- Established car and track graphics
- Modularized code to be compatible with previous path planning scripts
TO DO next week:
- Finish basic simulation graphics (i.e. turn right, stop)
- Test non-cooperative path planning algorithms on simulation

Discussed with team, TA and faculty member about change in design
- Project will be purely software-based
- Focus remains the same: showing the effect between cooperative vs non-cooperative autonomous driving
Worked on Statement of Work to go over parts of the project that are changing, being deleted or being added
Created simulation of figure-8 track with cars moving along the circle
- Below shows example of progress
- Yellow triangle denotes direction of car moving
- Modularized code to make it more dynamic
NEED TO DO NEXT:
- simulate cars turning right
- change speed based on path planning algorithms for non-cooperative case first
- possibly different scenarios (i.e. different tracks, different car interactions)
Progress is as expected, will continue to advance simulation

Met with team members to discuss the rescoping of our project and redistributed tasks.
Detailed project restructuring in the new design report
Created a high-level processing pipeline for cooperative path planning
- Identified useful cooperative signals for different track designs
TODO:
- Write starter code for cooperative v2v communication
- Integrate path planning code with graphic simulation

Design Report
- Finalized design report by clarifying design specifications such as tradeoffs, alternatives, metrics, etc.
- Organized what has been done and what still needs implementing/testing
Accomplishments
- Finalized path planning models/algorithms for cars (i.e. Intelligent Driver Model)
- From the construction of the car, figured out steering mechanism and movement control of vehicles
- Successfully constructed vehicle to move without glitches or power issues
- Tested communication latency between server to client (NodeMCU)
- Able to control one vehicle’s motors through Python server
TODO
- After parts come in, construct other vehicles and test communication across multiple vehicles
- Test latency between interaction with multiple clients
- Test path planning algorithms on cars for movement and speed
- Test camera detection on multiple vehicles running with tags
- Implement digital simulation of cars

Revised/finalized design report
Did more research on power issue and came up with a solution for the scope of our project
- After testing power on each part, decided to have different power supplies for NodeMCU and L298N driver so enough current would be available for the Wifi connection
- ordered more parts to test interacting with multiple vehicles
Was able to control vehicle’s motors through a Python server
- Connected NodeMCU to same WiFi network and controlled motion by byte commands
- Was able to do so without glitches or power issues
- Able to get an upper bound communication latency time
Progress is as expected
- After spring break when the parts come, will begin testing motion with multiple connections and multiple vehicles