At Vimaan.ai we build a wide range of products for warehouse inventory management. While our origins involve our indoor autonomous drone platform for cycle counting, we have since evolved to several computer vision based solutions for solving a host of other problems in the warehouse. ROS still remains a key part of our embedded capture systems' as well as server side pipelines' software architecture. In this talk, we would like to highlight the benefits as well as challenges we've faced with using ROS in the logistics industry even though we are not designing autonomous mobile robots per se.

We've engineered an advanced navigation system for ROS based mobile robots and maniplators, harnessing concepts of augmented reality (AR), Robot Operating System 2 (ROS 2), and the concept of Digital Twins. The current GUI used in the industry for controlling ROS robots are based on Web (using React JS, Foxglove) or native devices that work only on Tablets and PC. But our mission is to overcome navigation challenges by exploiting Google's ARCore platform, elevating user interaction and comprehension. Big Deep Tech Companies like Apple, Meta, Google, Jio are already getting ready for the future with AR/VR Glasses. And we are ready for this future by developing a universal application that can not only work on Mobile Phones, Tablets, PCs but also on AR and VR Devices and let you control and monitor ROS-based robots

The focus of this talk is to present a unified framework involving unit tests, integration tests, and systems tests with use of dockers and simulation-integrated CI/CD pipeline for generating deployment ready robot code with quality assurance in sight. These will be highlighted with a case study of a multi-robot navigation system.

The talk introduces a custom simulator built on Gazebo, enabling the creation of diverse outdoor simulations for UGV, UAV, USV, and UUV scenarios. It offers tunable effects like weather (fog, snow, dust, rain etc), dynamic sunlight, and ocean currents etc

People with whole-arm Amputations struggle to control prosthetic arms, and most prefer not to use them at all. Improving control and feedback interfaces might contribute – in the future – to making whole-arm prosthetics a practical reality. However, in the meantime, these devices remain inaccessible and expensive, for both users and researchers. We present a research platform for prototyping prosthetic control methods in a simulated environment. The system uses ROS to integrate marker-based motion capture, manual control, gaze tracking, intent estimation, and augmented reality visualisation. It can be adapted to various methods of manual input, and other kinds of wearable robots.

In this talk, I will share my experience building a web-first system interfacing with ROS 2 using Clojure. While it is uncommon to build ROS applications using Clojure, Clojure is an excellent language to build engineering-scale applications quickly. I will discuss my challenges during the implementation process, the choice of Clojure as the language, its ecosystem for ROS 2 (or the lack of it), and my innovative workarounds. Furthermore, I will introduce my open-source bridge library to implement a native interface for ROS 2 from within Clojure. This talk will benefit anyone implementing web interfaces for ROS.

Introduction of latest Zenoh bridges for ROS 1 and ROS 2 respectively, Brief about how Zenoh bridges improve responsiveness and minimize overhead by mapping ROS concepts directly to Zenoh, Introduction to how a teleoperation application can work seamlessly with ROS 1 and ROS 2 robots.

Simulink is the de-facto controller develoPMent tool in many industrial applications while ros_control is a real-time controller develoPMent framework for ROS. The motivation behind this project is to show how to fit a Simulink model into the ros_control framework. With the result of this project, you can auto-generate C++ code for a controller and package and deploy it as a ros_control plugin as to the ros_control framework

The work presents a simulation testbed developed for testing and demonstration of decentralized control algorithms designed for multi-agent systems. Aimed at bridging a gap between theory and practical deployment of such algorithms, this testbed provides a simulator with multi-agent systems having quadcopter agents. It is used to test the control algorithms designed assuming simple agent dynamics like a single or double integrator. This is based on the fact that under certain assumptions, quadcopter dynamics can be modeled as a double integrator system. A gazebo simulator with a physics engine as ODE (Open Dynamics Engine) is used for simulating the dynamics of the quadcopter. Robot Operating System is used to develop communication networks and motion control algorithms for the different agents in the simulation testbed. The performance of the test bed is analyzed by implementing linear control laws such as position control, leaderless consensus, leader-follower consensus, and non-linear control law for min-max time consensus

In this talk, we will present Beagle Flap, A flapping wing aerial robot with Linux-based AutoPilot. It enables the development of control and navigation algorithms for autonomous outdoor flight operations. Highlights include hardware overview, ArduPilot integration with MAVROS, development of Beagle Flap, custom add-on board design, and the implementation of DroneForce - a framework facilitating custom offboard controllers with Ardupilot and MAVROS. Additionally, insights are shared on leveraging open hardware projects such as BeagleBoard.org, can catalyze the Do-it-yourself designs for robotics research.

We present our work on autonomous row centering on standard orchards, which we have been working from our year by now. We will present how we are using ROS tools like gazebo, Rviz and Foxglove studio for testing and debugging our algorithm and also we will present what all the packages we use in developing our product and the challenges and opportunities in agriculture

The talk focuses on the integration of ROS (Robot Operating System) with custom or non-mainstream simulators and the benefits it offers. Simulations are crucial for verifying solutions before implementing them in robots that need to perform complex tasks autonomously. However, not all simulators are suitable for all projects. Projects like developing Gripper Algorithms, Algorithms for Driving with multiple actors, or Multi-Drone Algorithms require specific simulation components. Off-the-shelf simulators like Gazebo, Webots, or Carla are general 3D simulators and may not suffice for these projects. To address this, custom or non-mainstream simulators are necessary, but they may not be initially compatible with ROS. Manual integration with ROS enables the utilization of pre-existing algorithms, facilitates communication between components, and enables distributed processing, AMong other benefits.

Aquasee is an open source ROS package for underwater image enhancement. The aim is to democratise and widen the use of Autonomous Underwater Vehicles (AUV’s) and Remotely Operated Vehicles (ROV’s) for underwater exploration. The package provides a plug and play mechanism to enhance underwater images through hardware abstraction. The package provides rules to integrate any custom enhancement algorithm of your choice, making it developer friendly. The most iconic feature of the package is that it allows for water quality feedback from a TDS sensor. Total dissolved salts affect visibility because of dispersion and with additional feedback we model dispersion right into the enhancement suite.