74 research outputs found
A novel low-cost autonomous 3D LIDAR system
No full textThesis (M.S.) University of Alaska Fairbanks, 2018To aid in humanity's efforts to colonize alien worlds, NASA's Robotic Mining Competition pits universities against one another to design autonomous mining robots that can extract the materials necessary for producing oxygen, water, fuel, and infrastructure. To mine autonomously on the uneven terrain, the robot must be able to produce a 3D map of its surroundings and navigate around obstacles. However, sensors that can be used for 3D mapping are typically expensive, have high computational requirements, and/or are designed primarily for indoor use. This thesis describes the creation of a novel low-cost 3D mapping system utilizing a pair of rotating LIDAR sensors, attached to a mobile testing platform. Also, the use of this system for 3D obstacle detection and navigation is shown. Finally, the use of deep learning to improve the scanning efficiency of the sensors is investigated.Chapter 1. Introduction -- 1.1. Purpose -- 1.2. 3D sensors -- 1.2.1. Cameras -- 1.2.2. RGB-D Cameras -- 1.2.3. LIDAR -- 1.3. Overview of Work and Contributions -- 1.4. Multi-LIDAR and Rotating LIDAR Systems -- 1.5. Thesis Organization. Chapter 2. Hardware -- 2.1. Overview -- 2.2. Components -- 2.2.1. Revo Laser Distance Sensor -- 2.2.2. Dynamixel AX-12A Smart Serial Servo -- 2.2.3. Bosch BNO055 Inertial Measurement Unit -- 2.2.4. STM32F767ZI Microcontroller and LIDAR Interface Boards -- 2.2.5. Create 2 Programmable Mobile Robotic Platform -- 2.2.6. Acer C720 Chromebook and Genius Webcam -- 2.3. System Assembly -- 2.3.1. 3D LIDAR Module -- 2.3.2. Full Assembly. Chapter 3. Software -- 3.1. Robot Operating System -- 3.2. Frames of Reference -- 3.3. System Overview -- 3.4. Microcontroller Firmware -- 3.5. PC-Side Point Cloud Fusion -- 3.6. Localization System -- 3.6.1. Fusion of Wheel Odometry and IMU Data -- 3.6.2. ArUco Marker Localization -- 3.6.3. ROS Navigation Stack: Overview & Configuration -- 3.6.3.1. Costmaps -- 3.6.3.2. Path Planners. Chapter 4. System Performance -- 4.1. VS-LIDAR Characteristics -- 4.2. Odometry Tests -- 4.3. Stochastic Scan Dithering -- 4.4. Obstacle Detection Test -- 4.5. Navigation Tests -- 4.6. Detection of Black Obstacles -- 4.7. Performance in Sunlit Environments -- 4.8. Distance Measurement Comparison. Chapter 5. Case Study: Adaptive Scan Dithering -- 5.1. Introduction -- 5.2. Adaptive Scan Dithering Process Overview -- 5.3. Coverage Metrics -- 5.4. Reward Function -- 5.5. Network Configuration -- 5.6. Performance and Remarks. Chapter 6. Conclusions and Future Work -- 6.1. Conclusions -- 6.2. Future Work -- 6.3. Lessons Learned -- References
Robust control of geared and direct-drive robotic manipulators under parameter and model uncertainties
No full textThesis (M.S.) University of Alaska Fairbanks, 2005The major contribution of this thesis is the design and evaluation of a chattering-free sliding mode controller (SMC), which is a novel application for 2 degree-of-freedom (DOF) planar robot arms exposed to load variations. The performance of the SMC is evaluated in comparison to a proportional-derivative-plus (PD+) controller, as an example of nonlinear model-based controllers, as well as classical linear controllers, such as proportional-derivative (PD) and proportional-integral-derivative (PID). The performance of all four methods has been tested via realistic and detailed simulation models developed for both geared and direct-drive type 2-DOF planar robot arms. The model used in simulations reflects the dynamics of the arm, as well as the actuator dynamics and pulse width modulation (PWM) switching of the power converters. Simulations are performed under unknown load variations for both step and sinusoidal type reference joint trajectories. The results demonstrate that the chattering-free SMC provides increased accuracy and robustness than that of the other controllers and requires no prior knowledge of the system dynamic model and the load variation that the end-effector is subjected to. The results obtained could be extended to the control of a variety of geared and direct-drive type robotic configurations.Introduction -- Modeling of 2-DOF planar elbow manipulator -- Control of 2-DOF planar elbow manipulator -- Simulation results -- Conclusions and future work -- References -- Appendix
Sensorless sliding mode position control of induction motors using braided extended kalman filters
No full textIEEE International Symposium on Industrial Electronics -- JUN 04-07, 2007 -- Vigo, SPAINThis study is aimed at designing a sensorless sliding mode position control system for the rotor flux oriented Direct Vector Control (DVC) strategy of induction motors (IMs). For this purpose, a novel sliding mode controller (SMC) with reduced-chattering is designed for the control of the flux and angular position of the motor. All the states required for DVC in addition to the step-shaped load torque, stator resistance and rotor resistance are estimated using Braided EKF based observers. The performance of the new SMC is compared against a previously developed chattering-free SMC scheme. The simulation results demonstrate an improved robustness in the system response against parameter and load variations. It has also been demonstrated that the new Braided EKF technique used in the proposed sliding mode position control system also increases estimation accuracy of estimations when compared with chattering-free SMC under challenging variations of 100% in the load and stator & rotor resistance.IEEEGraduate School of University of Alaska FairbanksThis work was supported in part by the Graduate School of University of Alaska Fairbanks
- …
