Actuation
From Princeton Autonomous Vehicle Engineering
Goal: Design and build robust actuators and controllers to allow error-free actuation of brakes, throttle, steering and the gear shifter.
Team Lead: Talk to Josh for now.
Contents |
Status of Systems
Throttle
Lead: Mike
Status: in-progress. Specifications for completion are that the Throttle controller will respond to any angle (within a few degrees) more quickly than a human can.
There is no hardware component for this subproject. The electronic component is already complete (though a redesign for PCB may begin). The current goal is to translate the code to C++ utilizing the new Labjack classes. This involves rewriting the low-level analog outputs and the higher-level PID controller (recommendation is to utilize PID class, and to ignore program structure of previous iteration). P control is sufficient for this task. No further work is required or recommended.
The throttle is currently actuated electrically. LabJack1 digital output Cio3 controls the throttle relay. LabJack2 analog outputs Dac0 and Dac1 control the throttle signal.
Steering
Lead: Mike
Status: Complete with room for improvement. Specifications for completion are that the Steering controller will respond to any angle (within a few degrees) more quickly than a human can.
There is no hardware component for this subproject. The electronic component is already complete (though a redesign for PCB may begin). The current goal is to determine a more optimal set of control constants, perhaps adding a derivative component. This will likely require collecting data while driving the car for proper tuning.
The steering is currently actuated electronically. LabJack1 digital output Cio2 controls the steering relay. LabJack1 analog outputs Dac0 and Dac1 control the steering.
Brakes
Lead: Mike
Status: Partially Complete. The project is considered complete when they brakes can go from 1% position to 99% position within one second. This time frame and precision are not yet met, but the system still functions to a degree. Due to the minimum safe operating speed of the linear actuator, there is a maximum steady state deviation of 4% in the position. The system can go from ~1% to ~99% in about 2 seconds.
Right now the Victor (motor controller) is dangling by wires off the back of the Labjack Box and needs to be mounted more safely/permanently.
The old system is described below: The brakes are actuated electrically. There is a project currently to modify the brakes to be actuated mechanically. LabJack1 digital output Cio0 controlls the brakes relay. LabJack2 outputs two pwm signals to simulate the hydraulic and electrical brake sensors.
Gear Shifter
Lead: Josh
Status: Buttons (except L) are present and working. They're probably fragile though, so this control panel should be redone as soon as possible (but shouldn't interfere with other testing, since it does work...for now).
A completely new transmission interface will be designed that will completely remove the manual shifter, placing a linear actuator to switch gears and adding a user interface that uses buttons for gear selection. This project has the most to accomplish, and will thus require a three to four person team to achieve completion in a timely fashion.
There are several stages for this process:
Linear Actuator Mount - this involves moving the Escape's fusebox as currently designed, as it is in the way the slightly. There should then be enough space for the actuator and it's mount.
Modify Shifter - Remove the top component, cut it and connect the actuator. Make sure the actuator has long enough stroke at the point being connected to reach all gears (extremely important).
Create new panel interface - This can be hacked into place using the built in buttons temporarily to continue testing. However, someone will take this on as a design project and work on making an elegant, useful interface.
Connect wiring to Labjack (this will involve wiring up some relays - very simple, though may want to implement latch-on-latest-mechanism).
Code must be written to activate the relays appropriately.
The old system is described below: The gear shifter was actuated mechanically via a motor mounted under the engine. The motor actuates the gear shift cable directly by pulling and pushing it into the correct position. The motor is controlled via a Victor motor controller. The victor controller recieves a pwm signal and outputs the desired voltage to the motor. LabJack2 digital output Mio0 enables the transmission control line. LabJack2 timer 4 currently controls the pwm output to the victor.
Labjack Box Move
Lead: User:Esum
Status: it's in the back and wired, but the back panel is missing. All the cables that go into the Labjack Box should be routed through connectors on the back panel of the box, so they can be disconnected more easily (and so the box can be properly closed).
The electronics in the car currently have two issues. The mounting over the motor causes overheating issues in the summer that can cause fatal problems to crop up. The second issue is that the electronics are currently mounted to breadboards. The solutions are to move the electronics into the glove compartment and to solder them to PCBs. The glove compartment will be rebuilt, specialized to house the electronics.
Rear Shelf Rebuild
Lead: User:Herlands
Status: Shelf's built but batteries aren't installed yet (needs to be done soon). Labjack box was thrown in on top of this, so we should check if any other refinements are needed, too.
This shelf covers the inverter. The current complications are that it puts the rear power strip in a bad situation and it is aesthetically displeasing. The shelf is to be redesigned, with space put in for two extra car batteries (as an extra precaution for situations when the car may be unplugged and off).
Roof Rack Rebuild
Lead: Nathan
Mostly complete - the front mount needs to be stabilized. New camera mounts need to be built for this, and mounted sturdily (without shaking) on the roof rack.
Pimp the Vehicle
One thing to do is to make it look better than it currently does. A list of possible updates:
- Replace sensor housing with more streamlined covers.
- Black 26 inch rims.
- LCD TVs on both rear side windows.
- Grill
- Lots of random lighting for no reason whatsoever
Radar Gimbal
Info at Radar Gimbal.
Long Term Research
This area is a list of long term research projects that fall under actuation. They aren't meant to be implemented currently but hopefully will be completed and be migrated to Prospect 12 at some point in the future. They also provide interesting topics possibly for independent work.
Learning Controller
A PID controller that can learn its own constants and calibrate itself would be very beneficial. A learning controller would be able to adapt slowly overtime to adjust to different environmental conditions(a gimbal will move differently in the freezing cold and humid high heat situations.) It should be generic enough to be attached to any system that is currently on the car from throttle to gimbal controll.
Pan/Tilt Gimbal
Create a design for a gimbal which has two degrees of freedom. It should allow a sensor mount on it to point in any direction. This would be beneficial for providing true 360 degree awarness around the vehicle. It should mount sensors in the same manner as the current gimbal.
