Jens Christian Andersen

Orchard navigation using derivative free Kalman filtering

This paper describes the use of derivative free filters for mobile robot localization and navigation in an orchard. The localization algorithm fuses odometry and gyro measurements with line features representing the surrounding fruit trees of the orchard. The line features are created on basis of 2D laser scanner data by a least square algorithm. The three derivative free filters are compared to an EKF based localization method on a typical run covering four rows in the orchard. The Matlab® toolbox Kalmtool is used for easy switching between different filter implementations without the need for changing the base structure of the system.

MobotWare – A Plug-in Based Framework for Mobile Robots

Abstract This paper describes a plug-in based software framework developed at Automation and Control, DTU Electrical Engineering. The software has been used for education and research in mobile robotics for the last decade. Important design criteria have been real-time performance of the control level, easy integration of sensors, fast porting to new robots and core system stability and maintainability in an undisciplined programming environment. Real-time performance is assured by using RTAI-Linux; core stability is obtained by using plug-ins for user developed modules. The plug-in based module structure combined with inter-module communication based on TCP/IP sockets and human readable XML-protocol makes it easy to use the system on a wide range of hardware platforms, configurations and computer platform distributions. The framework has until now been interfaced to 7 different hardware platforms and has enabled many application i.e. robust navigation in an orchard with an autonomous tractor (Andersen,2010). Furthermore by providing a simple scripting robot control language the system also supports use by non-technicians.

Vision Assisted Laser Scanner Navigation for Autonomous Robots

This paper describes a navigation method based on road detection using both a laser scanner and a vision sensor. The method is to classify the surface in front of the robot into traversable segments (road) and obstacles using the laser scanner, this classifies the area just in front of the robot (2.5 m). The front looking camera is used to classify the road from this distance and forward, taking a seed area from the laser scanner data and from this estimate the outline of the visible part of the road. The method has been tested successfully on gravelled and asphalt roads in a national park environment.

Autonomous Tractor Navigation in Orchard - Diagnosis and Supervision for Enhanced Availability

Abstract Autonomous vehicles require a very high degree of availability and safety to become accepted by authorities and the public. Diagnosis and supervision are necessary means to achieve this. This paper investigates ways of using laser-scanner data to do localisation, and as a source of independent supervision, using expectation maximisation of laser-scanner output against uncertain map features. Analysis of system behaviours and their structure shows which redundant information is available to construct a supervisor. Tests on real life orchard data demonstrates the feasibility of the new approach.

Autonomous rule-based robot navigation in orchards

Abstract Orchard navigation using sensor-based localization and flexible mission management facilitates successful missions independent of the Global Positioning System (GPS). This is especially important while driving between tight tree rows where the GPS coverage is poor. This paper suggests localization based on an a priori map of the tree rows, a laser scanner based tree row detection, detecting both the row line and the row ends. The localization is combined with mission objectives in a rule-based inference interpreter. This rule-based mission handler combines the functional modules: localization, obstacle avoidance, path planning and drive control. The system is tested successfully using a Hako 20 kW tractor during autonomous missions in both cherry and apple orchards with mission length of up to 2.3 km including the headland turns.

Visual Navigation for Mobile Robots

Looking at the number of living creatures using vision as their main sensor one should expect that vision also would be the first choice for mobile robots considering nice features as low price, low power, non contact and high potential information contents. Unfortunately it has proven much more difficult to extract the information from vision than expected and still no commercial robot relies on vision as its main sensor. In spite of this several successful methods have been developed. This chapter presents a number of visual methods that has been experimentally verified: artificial visual landmarks, corridor following using vanishing point, and road following using terrain classification based on data fusion of laser scanner and vision.

CDIO AS THE EDUCATIONAL AND CULTURAL STRUCTURING ELEMENT IN THE DTU B.ENG. IN ELECTRONICS PROGRAMME

The aim of this paper is to describe how a CDIO based four semester study can be documented in such a way, that a homogeneous quality can be maintained over time. One purpose is to help new teachers to fully understand their role and obligations, not only in their particular course, but also as a part of the complex CDIO based education. The case used is the B.Eng. study in Electronics at the Technical University of Denmark (DTU) Implementing CDIO calls for many changes in the way that we build and document an program, having implemented CDIO at the B.Eng. program in electronics, it has been found that the normal public and internal course documentation platforms are insufficient to keep the large amount of information needed to describe the program as a whole, and the large amount of interaction between the individual courses, a master document describing the program has been developed to cover the first 4 semesters in the program, this paper is meant as an inspiration to others that might find this method beneficial. In todays modern and constantly changing society it must be expected that staff is constantly moving in and out between different research projects, while at the same time teaching courses at levels ranging from very advanced topics to introductory courses. In most cases a course will be given by the same teacher every semester, but for some courses (often the introductory courses) teachers change frequently. In this dynamic system the master document proposed helps in conveying crucial information from prior to new teachers, that otherwise could be lost in the teacher exchange process.

Derivative free Kalman filtering used for orchard navigation

In this paper the use of derivative free filters for mobile robot localisation is investigated. Three different filters are tested on real life data from an autonomous tractor running in an orchard environment. The localisation algorithm fuses odometry and gyro measurements with line features representing the surrounding fruit trees. The line features are created on basis of 2D laser scanner data by a least square algorithm. The Matlab® toolbox Kalmtool is used for easy switching between different filter implementations without the need for changing the base structure of the system.

Kalmtool used for laser scanner aided navigation in orchard

Abstract This paper concerns localisation of an autonomous tractor in an orchard environment, with the purpose of designing a localisation solution to be compared with GPS. The localisation is based on an estimate found by an extended Kalman filter, which fuses measurements from encoders and gyro with row measurements provided by a laser scanner. Kalmtool is used as a toolbox for developing the localisation algorithm. The result shows that the toolbox can be used successfully for dealing with localisation and sensor fusion.

An Adaptive Gain and Exposure Algorithm for Stereo Imaging

Imaging techniques are becoming more common as components of tractor perception systems for outdoor agricultural operations. When working outdoors, light conditions change continuously. The perception system should be able to continuously adapt and correct camera exposure parameters to obtain the best interpretation of the dynamically determined ROI (region of interest). We describe the development and testing of a procedure to automatically update gain and exposure parameter camera settings of a stereoscopic camera for use under dynamic light conditions. An algorithm based on an ideal ROI image histogram was developed and tested in an orchard and found to perform well under different orchard and light conditions.