The Need for High-Fidelity Robotics Sensor Models

Abstract

Simulations provide a safe, controlled setting for testing and are therefore ideal for rapidly developing and testing autonomous mobile robot behaviors. However, algorithms for mobile robots are notorious for transitioning poorly from simulations to fielded platforms. The difficulty can in part be attributed to the use of simplistic sensor models that do not recreate important phenomena that affect autonomous navigation. The differences between the output of simple sensor models and true sensors are highlighted using results from a field test exercise with the National Robotics Engineering Center's Crusher vehicle. The Crusher was manually driven through an area consisting of a mix of small vegetation, rocks, and hay bales. LIDAR sensor data was collected along the path traveled and used to construct a model of the area. LIDAR data were simulated using a simple point-intersection model for a second, independent path. Cost maps were generated by the Crusher autonomy system using both the real-world and simulated sensor data. The comparison of these cost maps shows consistencies on most solid, large geometry surfaces such as the ground, but discrepancies around vegetation indicate that higher fidelity models are required to truly capture the complex interactions of the sensors with complex objects.

Enabling Off-Road Autonomous Navigation-Simulation of LIDAR in Dense Vegetation

“…These empirical simulators can be quite realistic when predicting the performance of a specific autonomous system in a particular environment and are therefore useful in robotic development projects which already feature extensive field testing. However, these empirical simulators cannot fill the need for predictive, physics-based modeling [41].…”

Improving Sonar Sensor Fidelity in a Robot Simulator.

“…Mobile robots observe their environments through the medium of their sensors, so simulations should strive to model sensors to the highest level of fidelity [12]. However, modeling sensors using first principles is a difficult and time-intensive task, so most simulations make use of simplified, and often idealized, methods to reproduce sensor outputs [12].…”

“…However, modeling sensors using first principles is a difficult and time-intensive task, so most simulations make use of simplified, and often idealized, methods to reproduce sensor outputs [12]. For simulations to provide accurate and realistic training and testing the sensors used in the simulation should act as closely to real sensors as possible.…”

“…Despite the fact th a t sonar sensors have been around for decades surprisingly little re search has been done on how to model them in simulations [12].…”

“…Some researchers have used genetic algorithms and evolutionary programming [5] to make sensors work in their simulations; others have tried m athem ati cal models [20] of sensors. Simplistic, in the sense of being easy to program, models have been tried [12] but have not led to satisfactory results.…”