Gustavo M. Freitas

Robot Farmers: Autonomous Orchard Vehicles Help Tree Fruit Production

This article presents perception and navigation systems for a family of autonomous orchard vehicles. The systems are customized to enable safe and reliable driving in modern planting environments. The perception system is based on a global positioning system (GPS)-free sensor suite composed of a twodimensional (2-D) laser scanner, wheel and steering encoders, and algorithms that process the sensor data and output the vehicles location in the orchard and guidance commands for row following and turning. Localization is based on range data to premapped landmarks, currently one at the beginning and one at the end of each tree row. The navigation system takes as inputs the vehicles current location and guidance commands, plans trajectories for row following and turning, and drives the motors to achieve fully autonomous block coverage. The navigation system also includes an obstacle detection subsystem that prevents the vehicle from colliding with people, trees, and bins. To date, the vehicles sporting the perception and navigation infrastructure have traversed over 350 km in research and commercial orchards and nurseries in several U.S. states. Time trials showed that the autonomous orchard vehicles enable efficiency gains of up to 58% for fruit production tasks conducted on the top part of trees when compared with the same task performed on ladders. Anecdotal evidence collected from growers and workers indicates that replacing ladders with autonomous vehicles will make orchard work safer and more comfortable.

Kinematic reconfigurability of mobile robots on irregular terrains

This paper considers the active control problem of reconfigurable mobile robots on irregular terrain. A kinematic control strategy to improve robot mobility (stability and traction) is proposed. The proposed control is validated through numerical simulations and experimental tests using an amphibious wheel-legged robot, named Environmental Hybrid Robot, recently developed by Petrobras S.A. (Brazilian Petroleum Company) for environmental monitoring in the Amazon rain forest.

Kinematic control of constrained robotic systems

Este artigo considera o problema de controle de postura para sistemas roboticos com restricoes cinematicas. A ideia principal e considerar as restricoes cinematicas dos mecanismos a partir de suas equacoes estruturais, ao inves de usar explicitamente a equacao de restricao. Um estudo de caso para robos paralelos e robos cooperativos e discutido baseado nos conceitos de cinematica direta, cinematica diferencial, singularidades e controle cinematico. Resultados de simulacao, obtidos a partir de um mecanismo Four-Bar linkage, uma plataforma de Gough-Stewart planar e dois robos cooperativos, ilustram a aplicabilidade e versatilidade da metodologia proposta.

Multi-objective optimization for kinematic reconfiguration of mobile robots

The active control of reconfigurable mobile robots based on multi-objective optimization is considered. Different control strategies to improve the robot mobility are presented. Traction is considered in terms of the contact forces distribution among wheels, while stability is defined by a potential energy margin. Thus, considering the conflicting control objectives associated with these strategies, a multi-objective optimization approach is formulated to determine a set of optimal solutions which allows to establish a tradeoff solution. A planar robot is considered to illustrate the proposed multi-objective optimization approach.

A Robotics Framework for Planning the Offshore Robotizing Using Virtual Reality Techniques

The Oil & Gas industry has seen increasing costs of finding and extracting hydrocarbons, especially in remote locations, ultra-deep water reservoirs (400 m or deeper) or in hostile environments. Those new exploration frontiers have been increasing the production complexity and logistic costs. In such conditions, oil exploration feasibility depends on new technologies to optimize production efficiency. One possible solution to this challenge is to increase the degree of automation in production units. New design concepts also consider the use of robotic devices in such scenarios. In this paper we present a robotics framework, SimUEP-Robotics (Robotics Simulator for Stationary Production Units Unidades Estacionárias de Produção or UEPs, in Portuguese), aimed to enable planning the offshore platform robotizing using virtual reality techniques. The SimUEP-Robotics is based on ROS (Robot Operating System), a middleware for exchanging messages between different devices and processes that cooperate to accomplish a robotics task. SimUEP-Robotics is designed concerning the offshore requirements and is a flexible framework that allows the inclusion of new robots and devices in a virtual operation scenario. This capability enables the robotization and automation of offshore facilities that gradually evolve, starting from a complete virtual scenario towards a complete robotic system operating on a real platform, progressively including real devices. SimUEP-Robotics has an integrated Virtual Reality Engine (VR-Engine) specially tailored to provide realistic visualization of large offshore scene models in an immersive environment. The monitoring and management of remote operations of Stationary Production Units (SPU) is an activity that can also benefit by the usage of virtual reality scenarios due to the potential to reduce the complexity and difficulty in visualizing and validating simulations of operations performed by robots on a real SPU. The framework supports simultaneous simulation of multiple robots equipped with sensors and actuators like cameras, laser range finders and robotic manipulators. SimUEP-Robotics has also some specialized visualization tools like trajectory visualizer, ghostview robot animation, point-to-point measurement and a scenario editor that allows the user customize the target scenario accordingly. Through the use of those visualization tools it is possible, for example, to better understand the quality of the planned robot trajectory and propose new algorithms that can be further evaluated in the virtual environment. In conclusion, we argue that the validation process in an immersive virtual environment reduces risks and costs of real operation tests scenarios. SimUEP-Robotics has also an integrated Robotics-Simulator which is responsible for taking care of task planning and execution based on the information of the virtual scenario provided by the VR-Engine. To illustrate the effectiveness of the framework, different robotics applications were developed. One is an underwater application that calculates the whole dynamics of an operated ROV to simulate and test complex ROV operations in deep waters, like the connection of a flowline to a Christmas tree. The other one represents a topside offshore platform scenario where different virtual robots, derived from real mechanisms like Motoman DIA10, Puma 560, Seekur and others, operates. Results obtained on a pick and place task demonstrate the benefits of the proposed robotics framework for offshore applications.

A low-cost, practical localization system for agricultural vehicles

This paper addresses the refactoring of an agricultural vehicle localization system and its deployment and field-testing in apple orchards. The system enables affordable precision agriculture in tree fruit production by providing the vehicles position in the orchard without the use of expensive differential GPS. The localization methodology depends only on the wheel and steering encoders and the laser rangefinder already on the vehicle for row following, thus adding zero hardware cost to the overall setup. It employs an Extended Kalman Filter to integrate the information from the sensors, with the pose being predicted via encoder odometry and updated via point and line features detections. The objective of this paper is to describe the complete refactoring of the initial proof-of-concept localization system, with the goal of making it robust, modular and reusable. Field test results indicate that the final system has sufficient accuracy for deployment of autonomous vehicles in tree fruit orchards.

DESIGN, MODELING, AND CONTROL OF A WHEEL-LEGGED LOCOMOTION SYSTEM FOR THE ENVIRONMENTAL HYBRID ROBOT

Urucu, deep in the Amazon forest, is the largest in-land natural gas and oil producing site in Brazil. The oil and gas extracted in Urucu travel some 700 km in a pipeline through the jungle until they reach Manaus, where they are refined and consumed. Exploration of the site is conditioned on the Brazilian oil company Petrobras monitoring a 400 km section of the pipeline for any signs of water contamination. The Environmental Hybrid Robot (EHR) was conceived to automate this humongous monitoring task, currently executed in extremely small scale by scientists and locals on foot and small boats. This paper describes the design, analysis, and testing of the EHR’s second-generation locomotion mechanism, based on a two-DOF parallel mechanism. The new mechanism allows for independent control of wheel position and orientation relative to the terrain, thus significantly increasing the robot’s mobility. Additionally, it allows for control of the vehicle’s velocity-torque curve by adjusting the radius of curvature of the wheel at the point of contact with the soil. We present the forward and differential kinematic models of the new mechanism and, based on them, a differential kinematics-based position and orientation control method. We illustrate the methodology with experimental results.

Predictive Control of Actively Articulated Mobile Robots Crossing Irregular Terrains

Abstract This paper concerns the reconfiguration control of actively articulated mobile robots navigating through irregular terrains. Active mechanisms are able to accommodate for different operation conditions. The capability to influence mobility depends on the mechanism kinematic structure and the actuators velocity limitations. The robot mobility is evaluated considering ground clearance, orientation and stability. Control strategies are proposed considering the actuators bandwidth to compensate abrupt variations of the navigation trajectory and the driven terrain. The proposed solution consists on anticipating the command action by employing a predictive functional control method to adjust the robot for critical conditions faced during operation. Numerical simulations using field data, recorded while navigating on a natural terrain, are performed to verify the proposed strategies controlling a robot with two actuated DoF.

Modelagem de terrenos naturais através de malhas de triângulos

Nesse trabalho e apresentado um metodo de modelagem de terrenos naturais cobertos por vegetacao. A construcao do modelo e feita a partir da triangulacao de uma nuvem de pontos 3D, neste caso obtidos por um laser, gerando uma superficie 2 1/2 D. Testes realizados em um ambiente natural ilustram o funcionamento do metodo proposto atraves dos resultados obtidos. Os mapas gerados representam o terreno com detalhes, de acordo com a resolucao definida, que deve ser escolhida levando em conta o compromisso entre nivel de detalhamento e processamento computacional exigido.