Sebastián Sáenz
At the University of Texas at Austin
In the Cockrell School of Engineering
Junior ECE Student
Founder & Sole Member of Redux Systems
For Best Viewing Experience, Please Allow Tab to Occupy Full Screen, NOT compatible with IPad OS
Introducing R/AWP Rev. 1
The R/AWP - Remote/Autonomous Weapons Platform Revision One, A.K.A “Lil Guy” - is a light, versatile defense platform. Though currently outfitted with a .68 paintball marker, the R/AWP can platform a broad array of armaments with few modifications due to the simple yet effective frame system. The versatility of the platform is not the R/AWP’s only feature, it boasts a dual-mode remote control system, in which the operator can manually perform target acquisition from a safe distance with the wireless remote and onboard receiver controlled by nRF24L01 radio modules or allow the autonomous mode automatically acquire targets via artificial intelligence powered computer vision facilitated by OpenCV. The objects/entities the R/AWP can be trained to target are virtually limitless as it can be trained with ease.
Demo Reel
Meet Lil Guy.
FEATURES
FEATURES
Please wait a bit for the GIF to load or tap it :)
The chassis’ extreme durability and light weight is thanks to the fourteen lengths of 300mm & 500mm (six and eight respectively) 2020 series extruded aluminum and eight corner brackets that make up the chassis.
Extruded Aluminum Chassis
The majority of the turret itself is comprised of 3D printed components all of which are 100% designed and manufactured in house (by me, Sebastian) using Solidworks for CAD and a Prusa Mk3 i3 to create the parts. Every component is printed in PLA (polylactic acid) with high vertical wall counts and gyroid/hex infill. Each part boasts high tensile strength and durability while still remaining quite light.
3D Printed Construction
The R/AWP is a two-axis turret that boasts 360 degrees of freedom in the x plane and 70 degrees in the y plane. A 35kg/cm servo is responsible for each plane of movement while a third 35kg/cm servo actuates the trigger. Due to the nature in which servos are controlled, the R/AWP boasts precise, responsive handling that makes target acquisition a walk in the park.
Precise Servo Controls
The camera mounted under the paintball marker’s barrel is equipped with 850nm (infrared) LEDs and the chassis bears a powerful floodlight in the same frequency which permits the high/low light camera to operate in pitch darkness as easy as it would in daylight.
Low-Light Vision Capabilities
The R/AWP is controlled wirelessly with a handheld remote. Communication between the remote control and the turret itself is facilitated by the nRF24L01 chips which are abetted by signal boosters and voltage regulator chips to ensure a secure, continuous connection and low latency, as the data transmission rate is set to 240kbps which permits excellent responsivity without sacrificing battery life.
Remote Controlled
Targets can be acquired manually using a joystick onboard the remote control, but at the press of a button, the R/AWP can switch to an autonomous target acquisition mode in which an AI powered computer vision program facilitated by OpenCV takes over and R/AWP will automatically train its muzzle on a target and track it if it moves, the only necessary input being a trigger pull from the operator.
Autonomous Targeting Mode
All instruction processing, save for certain portions of the autonomous targeting system, takes place on two Arduinos. The remote has an Arduino Nano, which reads input data from the joystick and two onboard buttons, processes it, and sends it to the R/AWP which has an Arduino Uno onboard, which processes data from the Arduino and PC which are transferred wirelessly and via USB respectively.
Arduino Microcontrollers
To combat muzzle climb when firing, the R/AWP has an in-house designed compensator which redirect gas upwards to keep the muzzle on target when firing.
Climb Compensation System
Process Flow Diagram
Though less detailed than a circuit diagram, the process flow diagram provides an effective insight to the workings of the R/AWP. The Python script is constantly running on the PC, but essentially idles. Data is still sent to the Uno, but is discarded as it is not read in time. This idling process continues until the Nano sends word to the Uno to switch to the autonomous targeting mode, at which time the Uno will run the dedicated section of code for the autonomous mode in which the serial data from the PC is read, processed, and then sent to the servos. The camera feeds live video to the PC and is how the data the PC & Uno use to perform autonomous target acquisition is generated. In manual target acquisition ,mode, the Uno simply reads the data sent from the Nano and moves the servos accordingly to manually train the muzzle on a target.
