Developing software for large scale off-road robot applications is challenging and tedious due to cost, logistics, and rigor of field testing. High-fidelity sensor-realistic simulation can speed up the development process for perception and state estimation algorithms. We focus on Lidar simulation for robots operating in off-road environments. Lidars are integral sensors for robots, and Lidar simulation for off-road environments is particularly challenging due to the way Lidar rays interact with natural terrain such as vegetation. A hybrid geometric terrain representation has been shown to model Lidar observations well [1]. However, previous work has only been able to simulate a single, fixed scene, and the entire scene had to be precisely surveyed. In this work, we add semantic information to the hybrid geometric model. This allows us to extract terrain primitives, such as trees and shrubs, from data logs. Our approach uses these primitives to compose arbitrary scenes for Lidar simulation. We evaluate our simulator on a real-world environment of interest, and show that primitives derived using our approach generalize to new scenes.

There is growing interest in applying machine learning to motion planning. Potential applications are predicting an initial seed for trajectory optimization, predicting an effective heuristic for search based planning, and even predicting a planning algorithm for adaptive motion planning systems. In these situations, providing only a single prediction is unsatisfactory. It leads to many scenarios where the prediction suffers a high loss. In this paper, we advocate list prediction. Each predictor in a list focusses on different regions in the space of environments. This overcomes the shortcoming of a single predictor, and improves overall performance. A framework for list prediction, ConseqOpt, already exists. Our contribution is an extensive domain-specific treatment. We provide a rigorous and clear exposition of the procedure for training a list of predictors. We provide experimental results on a spectrum of motion planning applications. Each application contributes to understanding the behavior of list prediction. We observe that the benefit of list prediction over a single prediction is significant, irrespective of the application.

Sensor models, which specify the distribution of sensor observations, are a widely used and integral part of robotics algorithms. Observation distributions are commonly approximated by parametric models, which are limited in their expressiveness, and may require careful design to suit an application. In this paper, we propose nonparametric distribution regression as a procedure to model sensors. It is a data-driven procedure to predict distributions that makes few assumptions. We apply the procedure to model raw distributions from real sensors, and also demonstrate its utility to a mobile robot state estimation task. We show that nonparametric distribution regression adapts to characteristics in the training data, leading to realistic predictions. The same procedure competes favorably with baseline parametric models across applications. The results also help develop intuition for different sensor modeling situations. Our procedure is useful when distributions are inherently noisy, and sufficient data is available.

We study the problem of building a sensor model for the purpose of simulation. Our work is motivated by the potential impact of realistic simulators on the development cycle of software for real robots. The case is made for building models from approximate state information, relieving the burden of ground truth. Unlike calibration, where the goal is to identify and remove error from a signal, our aim to reproduce the signal in its entirety, including its error properties. Instead of physically modeling sensor behavior, a data-driven approach is taken. The implementation of our approach to simulate a simple but noisy laser rangefinder is described. Finally, approaches to validate the simulator are discussed. We compare not only raw sensor predictions, but also overall performance of algorithms on simulated versus real data.

We investigate the problem of a robot searching for an object. This requires reasoning about both perception and manipulation: certain objects are moved because the target may be hidden behind them and others are moved because they block the manipulator's access to other objects. We contribute a formulation of the object search by manipulation problem using visibility and accessibility relations between objects. We also propose a greedy algorithm and show that it is optimal under certain conditions. We propose a second algorithm which is optimal under all conditions. This algorithm takes advantage of the structure of the visibility and accessibility relations between objects to quickly generate optimal plans. Finally, we demonstrate an implementation of both algorithms on a real robot using a real object detection system.