Novel cable-driven robotic platforms and algorithms for environmental sensing applications

Vigilant monitoring of natural resources and biological phenomena is becoming increasingly important as population growth, security threats, and climate change focus attention on the integrity of our water and air. While static sensor networks have been effectively used to measure some phenomena, the high spatial variation of some variables of interest would require impractically dense deployments of sensors. Thus, conventional static sensor deployments may be enhanced, or even replaced, by mobile sensors capable of high-fidelity sensing of dynamic fields.;Sensor mobility may be achieved by equipping mobile devices with sensor payloads. However, rough terrain handicaps mobile agents in terrestrial applications, and wind or current may impede autonomous boats in aquatic settings. Moreover, localization and navigation in outdoor conditions is extremely difficult for conventional robotic platforms. Infrastructure-supported mobility, on the other hand, enables robotic devices to move freely within their workspace and can be leveraged for localization.;One of the first mobile robots geared toward environmental sensing was Networked Infomechanical System (NIMS), which used semi-permanent cabled infrastructure to enable sensing of various phenomena in a vertical planar transect. The subsequent NIMS-RD system vastly decreased deployment complexity, whereas NIMS-AQ enabled further improvements for aquatic scenarios. Significantly, these three systems were all limited to operation in a vertical plane. It was this limitation that spurred our development of cable-driven robotic systems capable of actuated sensing in three-dimensional environments. In this thesis, we describe two novel cable-driven parallel mechanisms designed specifically for such applications. NIMS3D is a rapidly deployable cable-array crane equipped with efficient in-field calibration techniques. NIMS for Planar Actuation (NIMS-PL) is a four-cable robot designed for planar actuation of a buoyant end-effector on a water surface. Three-dimensional operation is enabled for NIMS-PL by a motorized winch for vertical transport.;We have developed several algorithms to enable rapid deployability, in-field calibration, and energy-efficient autonomous operation of these systems. For NIMS3D, a rapid calibration scheme is presented by which untrained users can quickly and precisely determine the parameters of a deployment configuration. In addition, the use of nonlinear trajectories to reduce energy requirements is investigated. For NIMS-PL, we have developed two autonomous calibration methods by which the system can compute end-effector position with no prior knowledge of cable lengths. Coupled with a drift-detection algorithm, these methods eliminate long-term drift.;The existence of infinite tension arrangements in redundantly-actuated systems such as NIMS3D and NIMS-PL necessitates efficient solution of an underconstrained linear set of equations. We present low-power tension distribution methods for two classes of cable-driven robots: (1) Non-iterative methods for determining two-norm minimal distributions for fully-constrained systems; and (2) highly efficient iterative approaches for finding one-norm minimal distributions for configurations with redundant cables.;We present two methods of computing "safe" cable tensions which reduce operation at cable tension boundaries. The first method computes "Optimally Safe" tension distributions, which occur at the Chebyshev center of the feasible tension polyhedron. The second method of finding safe tension distributions computes the weighted-analytic center of the feasible tension polyhedron. W-center methods extend to any class of cable-driven parallel mechanism.;A set of example actuated sensing applications are presented, including topographical-mapping and light-intensity measurements performed by NIMS3D and a deployment of NIMS-PL in an aquatic environment. Furthermore, energy-efficient path-planning for actuated sensing is performed using NIMS-PL.