McGuinness, BenjaminShang, Ray2026-10-202026-10-202026https://hdl.handle.net/10289/18713Agricultural vehicles operating in orchard and pasture environments are required to perform frequent low-speed turning manoeuvres on uneven, grass-covered terrain - conditions under which steering strategy and vehicle geometry are major determinants of energy consumption. Despite a well-developed body of mathematical models predicting these effects, experimental validation has remained limited due to the absence of physical platforms that combine adjustable wheelbase, adjustable track width, and switchable steering strategies within a single vehicle. This thesis addresses that gap through the design, construction, and experimental evaluation of a fully electric, adjustable-geometry agricultural robot platform. The vehicle employs four independent in-wheel drive-steer modules, each integrating a brushless DC drive motor and a dedicated steering motor, mounted at the corners of a reconfigurable semi-monocoque chassis. Both wheelbase and track width can be independently adjusted across three values - 400 mm, 600 mm, and 800 mm - yielding a full 3×3 factorial matrix of nine geometric configurations. Six distinct steering modes are supported: two-wheel steering arc turn (R = 1.5 m), four-wheel steering arc turn (R = 1.5 m), skid-steer arc turn (R = 1.5 m), two-wheel point turn (R = 0), four-wheel point turn (R = 0), and skid-steer point turn (R = 0). Motor current is continuously logged at all eight drive and steer motor channels via a dedicated data acquisition system, enabling direct electrical energy quantification as E = Σ|I|·V·Δt. Controlled field experiments were conducted on two terrain conditions representative of New Zealand orchard and pasture environments: near-horizontal grass (mean inclination 0.216°) and a steep grass slope (mean inclination 17.618°). On flat terrain, all nine geometric configurations were tested across all six steering modes. On the slope, three full-width-track configurations (B = 800 mm) were evaluated for safety. Experimental results demonstrate that wheelbase L is the dominant geometric parameter governing turning energy consumption, with strong positive correlation across most steering modes (Pearson r = 0.80-0.90). Short-wheelbase configurations (L = 400 mm) consistently achieve 25-35% lower turning energy than long-wheelbase equivalents (L = 800 mm). The globally optimal configuration - B600×L400 and B800×L400 in four-wheel point-turn mode - achieves a minimum turning energy of approximately 707-710 J, representing a 31% reduction relative to the worst-performing configuration. Among all tested platforms, B800×L400 (track width 800 mm, wheelbase 400 mm, B/L = 2.00) produced the most outstanding overall performance across both flat and sloped terrain, exhibiting superior agility, smooth and rapid manoeuvre execution, and excellent directional stability, including on the 18° incline. Skid-steering modes uniformly incur the highest energy penalties across all geometric configurations (3-25× relative to the optimal equivalent mode), and a predictive L²/B parameter is shown to capture skid-steer energy with high accuracy (r = 0.86-0.98). Four-wheel steering consistently outperforms two-wheel and skid-steering strategies, and the arc four-wheel steering mode exhibits near-constant energy (~1,016 J, CV = 14%) across all nine configurations, making it the most geometrically robust steering choice. On slope terrain, arc-path manoeuvres incur substantially higher energy (+38% to +98%) relative to flat-ground equivalents due to gravitational work along the arc path, while pivot-type skid-steering energy is reduced by up to 22% due to decreased wheel-ground normal force. A systematic rearward shift in axle load distribution is identified during arc manoeuvres on slope, with rear-axle drive motors carrying up to 2.7× the energy of front-axle motors - a mechanically significant finding for rear-axle motor sizing in slope-capable platforms. Bivariate regression models of the form E = f(B, L) are derived for all six steering modes, achieving R² = 0.70-0.98 and providing practical engineering formulae for early-stage energy estimation from geometric parameters. The platform, its instrumentation methodology, and the experimental dataset collectively constitute a validated experimental foundation for steering-related energy research in electric agricultural vehicles. The findings directly inform design guidelines for energy-efficient small-scale agricultural robots, with particular relevance to orchard and vineyard applications in which turning energy represents a critical fraction of total operational energy.enAll items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.agricultural roboticsadjustable vehicle geometrywheelbase and track widthsteering strategiesturning energy consumptionenergy efficiency analysisenergy consumption patternsexperimental validationenergy prediction modelsThesis