Autonomous Quadcopter UAS P15230 Agenda Project Description / - - PowerPoint PPT Presentation

Autonomous Quadcopter UAS

P15230

Agenda

• Project Description / High Level Customer Needs / Eng

Specs

• Concept Summary
• System Architecture
• Design Summary
• System Testing Results
• Objective Project Evaluation: Success and Failure
• Opportunities/Suggestions for Future Work

Project Description

An autonomous Unmanned Aerial System (UAS) was built on an off-the-shelf quadcopter base. The quadcopter was designed in order to navigate an

• bstacle-filled course using various navigation techniques, including

simultaneous mapping and planning, object detection and avoidance, and basic facial recognition. The project was originally intended for use in the ImagineRIT “RIT Meets the Jetsons: Flying Cars” competition, and was designed as such. The project was later entered in the ARM Design Competition held during ImagineRIT.

Customer Requirements

Engineering Requirements

System Architecture: Overall

5 Architectural Components:

• 1. Flight Actualization
• 2. Mapping, Pathing,

and Object Detection

• 3. Localization
• 4. Facial Recognition
• 5. Safety

System Architecture - Electrical

• Power Distribution
• Signal Path

System Interconnection Printed Circuit Board

Electrical System-Power Management

• Battery
• a. Turnigy:4000mah
• Hall effect current sensor
• a. ACS 709

System interconnect

Power Management cont.

• Current and Voltage Reading
• a. Voltage Reading
• ADC on Teensy:13-bit
• Resistor Voltage Divider
• b. Current Reading
• ADC on Teensy:13-bit
• 3V max Resistor Voltage Divider

System Architecture - Embedded uC

• Interval Timer
• Pulse Position Library
• ISR Capture
• Analysis

Design Summary: Mechanical

• Frame:

○ DJI F450 Airframe Kit ■ Simple to replace components ■ Large Buildon space ■ Stable

• Propulsion

○ DJI E300 Tuned Propulsion Kit ■ 30A ESCs & 2212 920kV motors 9.4x4.3” props ■ 600 grams/axis rated max lift

• Craft lifts full load and has no issues
• vercoming initial lift at static thrust. Would be

cautious about adding much more weight to craft without adjusting propulsion system

• Rotor Shroud:

○ Designed to protect mechanical and electrical components of craft as well as bystanders and users.

• Leg Extensions:

○ 3D printed extensions raise craft for increased ground clearance. ○ Give guaranteed level take off ○ Allow for additional hardware mounting below lower plate of craft ■ Battery ■ Future add-ons

Design Summary: Mechanical cont.

• Current craft damage analysis

○ Shroud ■ collision with ceiling in lab caused corner to break off, but performed its job and protected both the craft and bystanders ○ Raspberry Pi 2 & Pi Cam ■ SD card slot on Pi 2 broke off in ceiling collision ■ Pi Cam 3D print bracket was damaged in collision, needs to be replaced

• Moving Forward (future safety/durability improvement)

○ Shroud weak points such as corners may be reinforced with bamboo skewers/bilateral tape ○ Protective covers can be produced for electrical hardware components that are more exposed

• n top of shroud (Raspberry Pi 2, Pi Cam, GY-

80, Teensy) ■ Con: Added weight ■ Alternative: Imbed more components into shroud foam

Shroud corner damage (repaired) Raspberry Pi 2 SD card slot Pi Cam Bracket

Design Summary: Software Class Diagram

Design Summary:Software State Chart

Design Summary: LIDAR Scanning Algorithm

System Testing: LIDAR Scanning

Algorithm

• Successfully tracks laser line pixel location indoors up to 30 ft away.
• Converts line pixels to real world locations, but needs more accuracy verification.
• Can convert detection to real world coordinates with simulated live compass heading and craft

location coordinates.

• Fixed-point C code formatted for full mapping system integration but never tested.

Hardware

• Pi NoIR camera successfully detects 980nm line laser indoors with IR pass filter. The sun and

bare incandescent light sources can cause issues.

• Line laser power output can be tuned down to safe ~1 mW levels at 3 ft away and still be visible

within 30ft.

• Raspberry Pi 2 runs standalone algorithm at about 5-15 fps, but data monitoring causes major

performance hit. Untested running full system integration.

Design Summary: Localization

Using three routers with known coordinates on the <x,y,z=0> grid, the Raspberry Pi is connected to each and a distance is calculated based on the signal strength of the router at that time. These distances are then used in geometric formulae to obtain the position of the craft.

Math used to derive equations for <x,y,z> coordinates

System Testing: Mapping & Path Planning

2D grid update mapping

• Left, Right, and back sonar sensors update the map if an object is detected

within a 1 meter range.Map update accounts for heading of the craft with initial calibrations at start of program.

• LIDAR integration untested
• Altitude determination of the craft varies by .5 meters from barometer

readings, and is limited by sonar range (<3m) Path Planning

• Path planning to multiple points on the same map is achieved in simulation
• Path planning with a dynamic map (updated by simulated sonar reading) is

achieved

• Craft not able to navigate with planned paths.

System Testing: Embedded uC

• Used I2C Master-Slave Arduino-to-Teensy

connection to emulate communication with Raspberry Pi

• Over Ride Switch - Confirmed Working

through RC UAS Test

• Autonomous Mode - Confirmed Working

though RC UAS Test

○ Further PID refinement necessary

System Testing - Localization

• WiFi-based triangulation proved to be too inaccurate for the precise

measurements needed for pathing algorithms.

• Calculated distances depended on signal strength, which did not vary

enough with distance to make accurate measurements.

• Could be used for larger-scale position finding, ie. for general location of

craft, but should no longer be considered for pinpoint measurements.

• For future versions of the project, RFID could be considered as a
• replacement. The triangulation method and final output would remain the

same, with only the distance calculations changing.

Project Evaluation

Current State: A quadcopter platform capable for research in autonomous

• flight. Hardware platform may interface

with any guidance system using i2C communication and defined packet structure, additionally full RC control is available to user. Main hardware components are off the shelf allowing for easy replacement.

Project Evaluation: ERs Status

S1:Complete Autonomous

• Craft does not fly completely autonomously, but has autonomous mode

implementable. S2: Take still images during flight

• Craft can not take still images, as the camera used continually throws

errors when used. S3: Store still images during flight

• Can not store still images, as there is no camera to record the images and

no dedicated storage space available on the craft. S4: Flys duration of function

• Craft has 10 minutes of continuous flight

Project Evaluation: ERs Status cont.

S5: Can fly with additional payload

• Craft can fly with additional payload, however this requirement was

no longer considered necessary for ImagineRIT. S6:Portable

• Can easily be carried by the user

S7: Onboard range sensing

• Sonar sensors are integrated into the onboard system. LIDAR

hasn’t been integrated into the onboard system. S8: Obstacle avoidance

• Craft isn’t tuned to avoid obstacles that it recognizes.

Project Evaluation: ERs Status cont.

S9: Safe for users and bystanders

• Craft can be taken into manual mode. Safe if experienced pilot.

S10: Stay within Budget

• Project total was $1125.14; this is $125.14 over budget.

S11: Facial Recognition

• Facial Detection implemented in MATLAB environment but not

integrated into onboard system. S12: Craft will not break on impact

• Craft take significant damage from drops of greater than 3 ft and

high speed collision with objects

Opportunities & Suggested Future Work

• 1. Finalize Flight Control

a. Tuning of Flight Controls (CC3D & Teensy PID) b. Face Tracking or Alternative Following UAS c. WiFi Positioning Integration To UAS For Indoor Flight d. GPS Positioning Integration To UAS For Outdoor Flight

• 2. Indoor Mapping (2D or 3D)

a. LIDAR Integration To UAS b. Use of RFID or similar system for more accurate coordinate position

• 3. Refine PCB for Late-Term Changes
• 4. Ergonomics

a. Battery b. Rx & Tx Binding Method