Files
Main/99 Work/My Experience/Experience at OneSec.md

135 lines
7.3 KiB
Markdown

---
title: Experience at OneSec
created_date: 2024-10-04
updated_date: 2024-10-04
aliases:
tags: CV
---
# Experience at OneSec
## Projects
- Electronic high level design of the drone
- make requirements list and come up with solutions to fulfil them
- redundancy was very important
- testing electrical parts (motors, escs, pdbs, carrierboards, antennas, gps chips, )
- learning electricity and how it works
- test battery cells on their specifications
- Wind Tunnel testing
- planning and prepare testing platforms (including arduino based flap control)
- Data acquisition (6-DOF force-torque sensor)
- Data analysis, conclude which Object behaves optimally for our usecase
- Understanding of Aerodynamics
- PCB Design
- Learn PCB design principles
- Kicad
- Full stack PCB design: requirements capture, schematic capture, pcb design, review (including external experts), finishing touches, ordering process, testing sequence
- Testing the pcbs: design testing protocol, build teststand, conduct testing, record data with oscilloscope and analyse data
- Lead electronics team and do review process of pcbs --> led to 90% success rate of functional pcbs in the first order
- Design the entire electrical hardware for drone from scratch using several PCBs, both flex PCBs as well as rigid ones.
- Reducing Wiring complexity by designing a wiring harness as a flexible pcb which drastically increases manufacturing efficiency
- Mechanical Design
- changed drone propulsion system design from 5 to 2 actuators by simplifying mechanical design
- fast iteration prototyping with 3D-printing, inserts, bushings, and test stand
- 3D printing (FDM, resin)
- Designing and use of a Propulsion System Teststand
- Use of raspberry pi, teensy (arduino like board) and several COTS sensors (current sensor, accelerometer, 6-DOF force torque sensor)
- Ability to run predefined testscripts or to steer the motor manually using RC
- auto upload to google drive
- Data Analysis Library for the teststand (python, pandas, bokeh, seaborn)
- Test and compare different propulsion systems for the drone that was developed
- Do system identification on drone
- Control System Design
- Implement the Control System in PX4 to make the novel propulsion system airborne.
- Host Flight log software on own server to speed up data analysis after every testflight (python, linux)
- Implement INDI control system from a MIT paper using PX4 and C++
- Flight Testing
- I've conducted hundreds of test flights to test mechanical parts, control system performances, sensor perfomances, autonomous software and others
- Hands-on & repair workshop
- I was repairing the drone after every crash using soldering, composite techniques, 3D-printing.
- I was responsible of the electrical assembly including soldering pcbs, cables, connectors and mounting everything in the drone
- Assisted in the manufacturing of the main body (carbon fiber, glass fiber)
- Helped monitoring and maintaining 3d printing farm and workshop (CNC, traditional machines, electrical workshop)
- Sensors and Actuators
- Test ESC-Motor combinations extensively and choosing adequate communication protocols
- Look at VESC in more detail
- integrate magnetic angle sensor into the PX4 Autopilot
- Integrate MIPI cameras into the linux system running on a rasperry pi
- Improve location accuracy by setting up RTK GNSS in PX4
- Enable LTE communication on raspberry pi by integrating a chip by quectel
- integrate and characterize Ark Flow: an optical flow module
- High Level Robotic Software
- Make sure that our custom AI-based algorithms run on the drone and steer it to the correct location using a ROS2 system and successfully run the first autonomous demos with the drone
- Rewrite the famous ROS2 based NAV2 library to work in 3D using custom map representations.
- Use TF2 to properly define the robot
- Use gazebo and px4 to simulate the robotic software stack
- Write docker containers that simulate the entire robot (ROS2 and PX4)
- Make sure two raspberry pis are synchronized using chrony daemon
- OSD Payload Controller
- use
## Learnings and Softskills
- Hiring:
- Went through hundreds of applicants and conducted technical interviews
- Teamlead
- leading the electrical hardware team as well as the robotic integration
- Challenges and benefits of startups
- money
- fast prototyping
- use of fiverr, upwork --> sometimes it needs to be done quickly without hiring someone completely
## Skills
- git
- C++
- Python
- Webdevelopment
- internal tools
- Linux
- Server maintanence
- Confluence and Documentation
- 3D Printing: FDM, Resin
- CNC machining: easy prototyping
- Solidworks
- PCB Design: schematic capture, component research, pcb design, pcb review process, pcb ordering process
---
## Electrical Engineering
In the beginning of my time at OneSec I was responsible for the design of the electrical hardware of the drone. One of the key design principles was redundancy, such that no single point of failure existed for safety critical features.
Before even designing the boards, I needed to conceptually plan the system and come up with a good solution in order to fulfil all requirements.
- Choose battery
- choose motors
- choose servos
- design power distribution board (PDB)
- buck converter circuits
- ideal diode solutions
- fuse system
- design carrierboard for different SoM computer chips (Google Coral, RB5 by Thundercomm and raspberry pi compute module 4)
- MIPI
- Ethernet
- PCIE --> Hailo
- Manage electrical engineering team
- Design reviews in order to guarantee quality and performance requirements
## Propulsion System Design
- mechanical and electrical design of a novel 4 DOF propulsion system
- 3D-printing prototyping
- mechanical design
- electrical design using magnetic sensors as well as COTS ESC
- development of the propulsion system firmware in C++
- implementation in the PX4 Autopilot software stack
- as well as a standalone solution from scratch
- Design and build of a propulsion system test stand that allows to measure several important key metrics such as propeller efficiency, propulsion system efficiency, torque actuation direction and strength. We used it to compare different propellers, to design the virtual Swashplate system and to make sure the torque actuation works as expected, comparison of BLDC motors, comparison of ESCs. The software consisted of the capturing system (scripting and manual control of the motor), which handled the motor control, the sensor logging as well as an automatic data upload to the cloud. Then there was the data analysis part, where the raw force torque sensor data needed to be processed in order to draw the desired conclusions.
The main hardware components used in the system:
- 6-DOF Force torque sensor
- current and voltage sensors
- magnetic angle sensor
- ESCs
- Teensy (Arduino like board)
- Raspberry Pi
## Data Analysis
Using python, pandas and several plotting libraries I was responsible for most of the engineering data analysis projects. This included:
- Wind Tunnel: aerodynamic testing of the drones body in order to maximize the cargo volume and lift while minimizing the aerodynamic drag.
- force torque sensor data
- normalize data to have valid comparisons between different models and wind tests
- rotate frames of reference in order to compare apples to apples
- Flight logs
## Flight Testing
-