Mission Disambiguation
Welcome to vehicle controls! Feel free to grab a coffee, sit down, and feel at home, this page is for you!
Building a Unified Controls Team
Section titled “Building a Unified Controls Team”Motivation
Section titled “Motivation”In its most abstract formulation, controls is concerned with a single problem: given a desired vehicle state, produce a set of inputs that most closely achieves that request while respecting a set of constraints over a given time horizon.
Framed this way, drivered and driverless controls emerge as seemingly identical problems with different sources of reference (in the case of DV path-planning’s trajectory, and in the case of EV the driver’s hands and feet), with everything downstream being identical. Given our driverless and electric vehicles share the same chassis, suspension, drivetrain, and firmware, in addition to the integrated direction FSAE is going, it is only natural to unify the two teams into a single effort that pools our analysis resources, knowledge base, and engineering talent! Of course this is a huge simplification of an extremely difficult engineering problem, but join along for the ride!
Vehicle controls owns the following responsibilities on the team:
Low-level vehicle controller that interfaces with motor controllers, performs active torque vectoring, and handles regen balance. Works closely with Firmware.
- Inputs
- Accepts driver and driverless control inputs
- Steering wheel
- Gas and brake pedal positions
- Reads in vehicle state
- velocity
- acceleration
- orientation
- heading
- Accepts driver and driverless control inputs
- Outputs
- 4-wheel torques
Projects: General Vehicle Controller R&D, Regenerative Braking R&D, Monorepo QoL
High-level driverless trajectory planner; responsible for taking a spline and environment state and outputting a control action.
Projects: Performance & Profiling, Algorithm R&D, Driverless QoL (good for beginners!)
Testing harnesses and simulators used to validate EV, DV, and unified controls code before it reaches the car. Works closely with DV Software Architecture.
Projects: Carmaker HiL/SiL, DV Controls Simulator
Vehicle models, system identification, and state projection; quantifies how the car actually behaves and how far our sim is from reality. Works closely with Vehicle Dynamics.
Projects: System Identification & Tooling, Vehicle Modeling R&D, State Projection QoL (good for beginners!)
Controls Goals for 27x
Section titled “Controls Goals for 27x”High-level Focus Areas
Section titled “High-level Focus Areas”- Create software that helps us better understand the vehicle state during and post vehicle operation; construct ways to quantify the effectiveness of the software we write.
- Design and implement intuitive tuning interfaces that directly correlate to vehicle performance characteristics.
- Build out robust suite of testing platforms to validate every line of code we deploy to the vehicle.
Team Structure
Section titled “Team Structure”We will be adopting the DV ticket system for managing controls tasks next year. This will be used during the Summer/Fall, but I will make changes if necessary when it seems like this structure is not sufficient for our needs. Feedback on how you like this new ticketing structure will be very much appreciated!
Ticket Based Tasking
Section titled “Ticket Based Tasking”All projects will be broken down into incremental task-driven tickets that will be assigned to controls members; the hope is that a ticket-driven approach to assigning work will help relieve the burden of members having to pace their own projects, it also makes it easier for me to check-in on and track deliverables.
System Leads
Section titled “System Leads”Controls will still have system leads next year but they will function more like system owners. This allows for the flexibility of working beyond your system, and should allow for system leads to accumulate more knowledge across all domains as they work on tickets that span the range of controls.
