Enhancing Robotic Precision in Construction: A Modular Factor Graph-Based Framework to Deflection and Backlash Compensation Using High-Accuracy Accelerometers

TL;DR

This work tackles precise end-effector localization for construction robots with long kinematic chains by introducing a modular factor-graph framework that fuses deflection and backlash models with high-accuracy accelerometers. It formulates a stationing routine and a tight integration of physics-informed deflection models with accelerometer data to estimate the kinematic-state, sensor biases, and calibration parameters in a probabilistic setting. Key contributions include a DeflectionFactor and DeflectionPriorFactor, a publicly released 3090-sample dataset, and demonstrated mm-scale improvements: a 50% reduction in the 95th percentile xy error over the state-of-the-art and a 31% gain when incorporating base tilt compensation, across diverse site disturbances. The approach offers robust, modular, and scalable improvements to autonomous construction tasks where continuous external tracking is impractical.

Abstract

Accurate positioning is crucial in the construction industry, where labor shortages highlight the need for automation. Robotic systems with long kinematic chains are required to reach complex workspaces, including floors, walls, and ceilings. These requirements significantly impact positioning accuracy due to effects such as deflection and backlash in various parts along the kinematic chain. In this work, we introduce a novel approach that integrates deflection and backlash compensation models with high-accuracy accelerometers, significantly enhancing position accuracy. Our method employs a modular framework based on a factor graph formulation to estimate the state of the kinematic chain, leveraging acceleration measurements to inform the model. Extensive testing on publicly released datasets, reflecting real-world construction disturbances, demonstrates the advantages of our approach. The proposed method reduces the $95\%$ error threshold in the xy-plane by $50\%$ compared to the state-of-the-art Virtual Joint Method, and by $31\%$ when incorporating base tilt compensation.

Figures (8)

• Figure 1: A prototype of the Hilti JaiBot, together with a total station, used in this study to evaluate our model. The individual components of the system are annotated in the image.
• Figure 2: Factor graph formulation of the model. We use the DeflectionPriorFactor only during stationing and the PositionFactor only if $\pmb{\mathcal{M}}_\text{ex}$ available.
• Figure 3: A 2D visualization of the deflection model. The model is composed of a compliance part modeling the base-environment interaction and linear springs with backlash for each column joint.
• Figure 4: Waypoints, through which the robot navigates for every lifting column height using the standard solution space.
• Figure 5: Setups of the robot and its environment for the respective datasets.