Marc Freese

Bio: Marc Freese is an academic researcher from Tokyo Institute of Technology. The author has contributed to research in topics: Robot & Demining. The author has an hindex of 7, co-authored 16 publications receiving 1303 citations.

V-REP: A versatile and scalable robot simulation framework

Abstract: From exploring planets to cleaning homes, the reach and versatility of robotics is vast. The integration of actuation, sensing and control makes robotics systems powerful, but complicates their simulation. This paper introduces a versatile, scalable, yet powerful general-purpose robot simulation framework called V-REP. The paper discusses the utility of a portable and flexible simulation framework that allows for direct incorporation of various control techniques. This renders simulations and simulation models more accessible to a general-public, by reducing the simulation model deployment complexity. It also increases productivity by offering built-in and ready-to-use functionalities, as well as a multitude of programming approaches. This allows for a multitude of applications including rapid algorithm development, system verification, rapid prototyping, and deployment for cases such as safety/remote monitoring, training and education, hardware control, and factory automation simulation.

Virtual robot experimentation platform V-REP: a versatile 3D robot simulator

Abstract: From exploring planets to cleaning homes, the reach and versatility of robotics is vast. The integration of actuation, sensing and control makes robotics systems powerful, but complicates their simulation. This paper introduces a modular and decentralized architecture for robotics simulation. In contrast to centralized approaches, this balances functionality, provides more diversity, and simplifies connectivity between (independent) calculation modules. As the Virtual Robot Experimentation Platform (V-REP) demonstrates, this gives a smallfootprint 3D robot simulator that concurrently simulates control, actuation, sensing and monitoring. Its distributed and modular approach are ideal for complex scenarios in which a diversity of sensors and actuators operate asynchronously with various rates and characteristics. This allows for versatile prototyping applications including systems verification, safety/remote monitoring, rapid algorithm development, and factory automation simulation.

Teleoperated Buggy Vehicle and Weight Balanced Arm for Mechanization of Mine Detection and Clearance Tasks

Abstract: The development of a practical mobile robot system for mechanization of humanitarian demining tasks such as sensing/verification and/or clearance of anti-personnel landmines is close to be completed. The system proposed by Hirose group at Tokyo Tech is a simple but effective solution that consists of a pantographic weight-balanced manipulator mounted on top of a mobile platform. A practical and versatile mobile platform has been successfully realized by converting a commercial 4- wheel buggy for remote-operation. Nonetheless, it still preserves the characteristics of the original buggy, and can conveniently serve as a commuting vehicle for the human worker as well. On the other hand, the manipulator mounted on the buggy is a complete new type of manipulator especially designed for the landmine detection and removal tasks. It adopts a new weight- compensated configuration consisting of 4 links and 5 nodes pantographic linkage. This manipulator can be equipped with many types of sensors to detect landmines and also has the capability to equip grass cutter, prodders and other tools if so required. For the short term in particular, this research is focused on the sensing tasks. For this purpose, mine sensors composed of metal detector and ground penetrating radar has been tested. Furthermore, automatic topographical mapping generation and scanning of the terrain has also been implemented using stereo-vision cameras mounted on the manipulator link. This automation relieves the human-operators from the burden of watching all the scanning process which can be long and demanding. Nonetheless, the operator can switch to manual scanning and teleoperate the manipulator whenever it is necessary. In this paper the authors explain the system main concepts and give details of its main composing parts. The validity and usefulness of the proposed system are demonstrated by real experimental results.

Endpoint Vibration Control of a Mobile Mine-Detecting Robotic Manipulator

Abstract: The procedure for humanitarian land mine removal varies greatly. However, one central element is detecting and marking suspected mine locations using sensors such as trained dogs and metal detectors. This step is expensive, tedious, and dangerous. In order to improve the land mine detection process, a reliable, low-cost robotic manipulation system has been constructed. The system consists of long-reach manipulator mounted on a commercially available All-Terrain Vehicle. The system is capable of autonomously scanning for mines with different types of sensors. An electromagnetic induction metal detector is used in conjunction with ground-penetrating radar to detect metal and larger heterogeneities in the ground, i.e. mine detonators and mine casings, respectively. This sensor combination greatly reduces the number of false alarms over the traditional single-sensor approach. Given the long reach of the arm, the endpoint is susceptible to unwanted oscillations that corrupt the scanning information. This paper describes a control system that utilizes input shaping to improve endpoint tracking. The result is a system that generates high-quality sensor information for precise mine identification.

Robotics-assisted demining with Gryphon

Abstract: The threat and consequences of landmines have led to a multitude of alternative research activities in the field of demining. While mine sensor-focused research has been intensive, there has been relatively less attention given to the problem of automating the detection and removal procedure. Understandably, autonomous robot operation and interaction in unstructured field environments are difficult. This paper addresses this by presenting a robot meant to assist humanitarian demining by providing a cheap, fast, reliable and safe alternative to human deminers risking their lives on a daily basis. The robot, named Gryphon, is able to autonomously scan 2 m2 at a time with any type of mine sensor payload. It then presents acquired sensor images to the operator who selects which spots need further investigation or prodding. The robot then appropriately marks the terrain with paint or marking plates. Gryphon has been extensively field tested in Japan, Croatia and Cambodia.

Computer vision : a modern approach = 计算机视觉 : 一种现代的方法

Abstract: From the Publisher: The accessible presentation of this book gives both a general view of the entire computer vision enterprise and also offers sufficient detail to be able to build useful applications. Users learn techniques that have proven to be useful by first-hand experience and a wide range of mathematical methods. A CD-ROM with every copy of the text contains source code for programming practice, color images, and illustrative movies. Comprehensive and up-to-date, this book includes essential topics that either reflect practical significance or are of theoretical importance. Topics are discussed in substantial and increasing depth. Application surveys describe numerous important application areas such as image based rendering and digital libraries. Many important algorithms broken down and illustrated in pseudo code. Appropriate for use by engineers as a comprehensive reference to the computer vision enterprise.

V-REP: A versatile and scalable robot simulation framework

Abstract: From exploring planets to cleaning homes, the reach and versatility of robotics is vast. The integration of actuation, sensing and control makes robotics systems powerful, but complicates their simulation. This paper introduces a versatile, scalable, yet powerful general-purpose robot simulation framework called V-REP. The paper discusses the utility of a portable and flexible simulation framework that allows for direct incorporation of various control techniques. This renders simulations and simulation models more accessible to a general-public, by reducing the simulation model deployment complexity. It also increases productivity by offering built-in and ready-to-use functionalities, as well as a multitude of programming approaches. This allows for a multitude of applications including rapid algorithm development, system verification, rapid prototyping, and deployment for cases such as safety/remote monitoring, training and education, hardware control, and factory automation simulation.

Virtual-to-real deep reinforcement learning: Continuous control of mobile robots for mapless navigation

Abstract: We present a learning-based mapless motion planner by taking the sparse 10-dimensional range findings and the target position with respect to the mobile robot coordinate frame as input and the continuous steering commands as output. Traditional motion planners for mobile ground robots with a laser range sensor mostly depend on the obstacle map of the navigation environment where both the highly precise laser sensor and the obstacle map building work of the environment are indispensable. We show that, through an asynchronous deep reinforcement learning method, a mapless motion planner can be trained end-to-end without any manually designed features and prior demonstrations. The trained planner can be directly applied in unseen virtual and real environments. The experiments show that the proposed mapless motion planner can navigate the nonholonomic mobile robot to the desired targets without colliding with any obstacles.

Learning Synergies Between Pushing and Grasping with Self-Supervised Deep Reinforcement Learning

Abstract: Skilled robotic manipulation benefits from complex synergies between non-prehensile (e.g. pushing) and prehensile (e.g. grasping) actions: pushing can help rearrange cluttered objects to make space for arms and fingers; likewise, grasping can help displace objects to make pushing movements more precise and collision-free. In this work, we demonstrate that it is possible to discover and learn these synergies from scratch through model-free deep reinforcement learning. Our method involves training two fully convolutional networks that map from visual observations to actions: one infers the utility of pushes for a dense pixel-wise sampling of end-effector orientations and locations, while the other does the same for grasping. Both networks are trained jointly in a Q-learning framework and are entirely self-supervised by trial and error, where rewards are provided from successful grasps. In this way, our policy learns pushing motions that enable future grasps, while learning grasps that can leverage past pushes. During picking experiments in both simulation and real-world scenarios, we find that our system quickly learns complex behaviors even amid challenging cases of tightly packed clutter, and achieves better grasping success rates and picking efficiencies than baseline alternatives after a few hours of training. We further demonstrate that our method is capable of generalizing to novel objects. Qualitative results (videos), code, pre-trained models, and simulation environments are available at http://vpg.cs.princeton.edu