In this phase in Blue Ocean Robotics I reduced my time as a developer to focus on the planning and design of the UVD Model C robot.Integrate multiple domain teams (Hardware, Mechanics, Autonomy and UVD Core) to develop the UVD Model C features.I lead the UVD Core team defining tasks and running scrumban sessions.Integrate with management team to report and receive priorities.Participate in the the engineering change process to organize the development of the ECs and pass to other sectors in the company like sourcing, manufacturing.Acted more as third tier support and wrote SOPs to help Customer Service to troubleshoot problems on the field.On the software release process and integration with Customer Service, I'm the front door to evaluate possible bugs to analyse, then confirm/reject them, and/or assign to the correct team.In the new role my focus on the features has shifted from just development:- Write technical requirements with UX team- Write design and architect documents- Define Jira tasks and assign to the appropriate team leadersWorked on the development of the features:- Audio Feedback for the users to be aware of what the robot is executing- Maintenance Mode for the robot to back away from obstacles that are too close- Fleet Management solution for users to see the disinfection data on the cloud. ROS nodes interface with a RealmDB (nodejs) inside the robot that synchronizes with MongoDB in the Cloud- Charging Schedule to make sure the batteries are fully charged when the operators need to use the robot. Used a gradient optimization method to predict charging/discharging behaviours.- Upgrade of Ubuntu and ROS version of the robots on the field

UVD RobotsThe UVD Robot Model C are mobile differential autonomous robots that uses UV light to disinfect surfaces reduce infection in hospitals. The robot is controlled by a tablet that connects to the robot's WiFi. The user can map the environment, set disinfection positions and run disinfection in rooms. The robot uses LIDARs and depth cameras to navigate autonomous.The robot software is implemented with ROS 1. It uses a local database to store information about the map and disinfection. In respect to the ROS autonomy stack, it uses a modified autonomy stack to handle the costmaps and planners, and the cognition is implemented using Behaviour Trees. The robot serves a Webapp for the tablets.The robot has a TÜV certification stating that the robot is safe to use around humans. The software development was going through QA team that would test the robot with each software version to ensure new features are implemented correctly and no regression was introduced.My development work was focused was on hardware interfaces and cognition stack. Until the robot was stable, a big part of the time was used in bug fixes.I worked in multiple new features and improved existing ones:- Events monitor from system info and ROS messages (warning, errors)- Error handling, recovery and logging- Interface with I/Os (relays, buttons, PLC)- Wireless charger/battery interfaces- Charging logic (State Machine)- Disinfection workflow- Disinfection interruption logic and automatic recovery- Integration between Backend and Frontend through rosbridge- Robot power up/down process (State Machine)- Infrastructure/Automation - Configuration of the developers and robot's computers using Ansible - Pipelines to build individual ROS packages and run unit/integration tests, and deployment - Rosdistro and bundle of the softwareBesides development, I help transition the robot into Manufacturing and later acted as third tier support for Customer Support for robots in the field.

Advisor for the consultant contracts

Prospects of R&D projects

Project ARIEL (Repsol, ANP, COPPETEC, GSCAR, Tidewise, Farol)ARIEL system is autonomous system for oil spill detection on the ocean utilizing a commercial drone and an autonomous boat being developed by GSCAR and Tidewise.Integration of the systems is key factor on the project, and the autonomous landing of the drone on the boat is the biggest challenge.GSCAR will work on the following subjects: state machines, trajectory planning, control algorithm, visual marker detection and Kalman filter. The robot would have a DJI M600 with a Raspaberry Pi and PTZ camera.I left the project within 1/4 of the project time, but the project was successful and won first place in Prêmio ANP 2020 (Brazil's National Agency of Petroleum).

Project HEADS (Repsol, ANP, COPPETEC, GSCAR, 13 Robotics, Tecgraf, Indra)HEADS is a system capable of detecting oil spill at sea within 2 km utilizing radar and infrared camera in a static environment.For the system to work on a FPSO or oil platform, a ISP (Inertially Stabilized Platform) was developed to stabilize the thermal camera on a point on the sea.The prototype was mounted on top of a Stewart Platform that would generate movements similar to the vehicle that would carry the HEADS.GSCAR developed the control algorithm for the ISP while 13 Robotics developed the electronics, mechanics and hardware integration.The software was implemented in Rock, using computed torque in a real time control loop to reach high precision.The system was developed and tested on the Gazebo simulator.The simulation environment and the system share the same code.I and colleagues in GSCAR were tasked with the simulated environment, interface with the Stewart platform, modeling the problem, designing the control algorithm, implementing the control algorithm in C++ and ROCK, testing and tuning the control with the prototype.We implemented a Line Of Sight control using a cascade control of PID and computed torque taking into account the dynamics of the 3DOF ISP and the movement of the base, plus a compensation for the joint friction.

Project DORIS (Petrobrás, Statoil, ANP, COPPETEC, GSCAR, SMT)DORIS is a mobile robot for remote supervision, diagnostic and data acquisition in oil platforms guided by rails developed from scratch in a R&D project.It can carry sensors fit for each inspection environment and can process audio and video anomalies. It has a lightweight manipulator with a camera and pressure sensor to interact with the environment.I was responsible for the development and management of the robot software on the embedded electronics (computer with PC104 bus) and on the base (notebook).The software does the interface between hardware (motor drivers, LIDAR, câmeras, microphones), communication layer (TCP/IP, CAN/CANopen, USB) and the GUI.The framework ROS (Robot Operating System) was used as the base of all the software.The GUI developed for the project focused on the end user experience and had generic tools.The project won first place in the Prêmio ANP 2017 (Brazil's National Agency of Petroleum).

Project DIANE (Electronics, software and control)A EOD (Explosive Ordnance Disposal) robot constituted of track for locomotion, arms to climb and descend stairs, a SCHUNK robotic manipulator of 3 DOF to manipulate the explosive, cameras for operation, EPOS controllers and a embedded PC with QNX.I worked on the electronics and software.The software was developed in C++ and used MFC library for the GUI, pthread (parallel programming), socket (TCP/IP network) and NTCAN (library for the CAN/CANopen networks).

Project SINAPEThe activity in CEPEL was constituted on the maintenance of software SINAPE (Integrated System of Support of Perturbation Analysis) utilizing C++ and MFC library.The program provides tools to analyse perturbations on the power lines that could potentially damage the network.