19 June 2019 in IEEE Robotics and Automation Letters
28 January 2019 in IEEE Robotics and Automation Letters
Legged robots have the potential to traverse diverse and rugged terrain. To find a safe and efficient navigation path and to carefully select individual footholds, it is useful to be able to predict properties of the terrain ahead of the robot. In this work, we propose a method to collect data from robot-terrain interaction and associate it to images. Using sparse data acquired in teleoperation experiments with a quadrupedal robot, we train a neural network to generate a dense prediction of the terrain properties in front of the robot. To generate training data, we project the foothold positions from the robot trajectory into on-board camera images. We then attach labels to these footholds by identifying the dominant features of the force-torque signal measured with sensorized feet. We show that data collected in this fashion can be used to train a convolutional network for terrain property prediction as well as weakly supervised semantic segmentation. Finally, we show that the predicted terrain properties can be used for autonomous navigation of the ANYmal quadruped robot.
10 September 2018 in IEEE Robotics and Automation Letters
This letter presents a case study of a learning-based approach for target-driven mapless navigation. The underlying navigation model is an end-to-end neural network, which is trained using a combination of expert demonstrations, imitation learning (IL) and reinforcement learning (RL). While RL and IL suffer from a large sample complexity and the distribution mismatch problem, respectively, we show that leveraging prior expert demonstrations for pretraining can reduce the training time to reach at least the same level of the performance compared to plain RL by a factor of 5. We present a thorough evaluation of different combinations of expert demonstrations, different RL algorithms, and reward functions, both in simulation and on a real robotic platform. Our results show that the final model outperforms both standalone approaches in the amount of successful navigation tasks. In addition, the RL reward function can be significantly simplified when using pretraining, e.g., by using a sparse reward only. The learned navigation policy is able to generalize to unseen and real-world environments.
5 February 2018 in IEEE Robotics and Automation Letters
Global registration of multiview robot data is a challenging task. Appearance-based global localization approaches often fail under drastic view-point changes, as representations have limited view-point invariance. This letter is based on the idea that human-made environments contain rich semantics that can be used to disambiguate global localization. Here, we present X-View, a multiview semantic global localization system. X-View leverages semantic graph descriptor matching for global localization, enabling localization under drastically different view-points. While the approach is general in terms of the semantic input data, we present and evaluate an implementation on visual data. We demonstrate the system in experiments on the publicly available SYNTHIA dataset, on a realistic urban dataset recorded with a simulator, and on real-world StreetView data. Our findings show that X-View is able to globally localize aerial-to-ground, and ground-to-ground robot data of drastically different view-points. Our approach achieves an accuracy of up to 85% on global localizations in the multiview case, while the benchmarked baseline appearance-based methods reach up to 75%.
18 November 2019 in IEEE Robotics and Automation Letters
Globally consistent dense maps are a key requirement for long-term robot navigation in complex environments. While previous works have addressed the challenges of dense mapping and global consistency, most require more computational resources than may be available on-board small robots. We propose a framework that creates globally consistent volumetric maps on a CPU and is lightweight enough to run on computationally constrained platforms. Our approach represents the environment as a collection of overlapping SDF submaps, and maintains global consistency by computing an optimal alignment of the submap collection. By exploiting the underlying SDF representation, we generate correspondence-free constraints between submap pairs that are computationally efficient enough to optimize the global problem each time a new submap is added. We deploy the proposed system on a hexacopter MAV with an Intel i7-8650U CPU in two realistic scenarios: mapping a large-scale area using a 3D LiDAR, and mapping an industrial space using a RGB-D camera. In the large-scale outdoor experiments, the system optimizes a 120x80m map in less than 4s and produces absolute trajectory RMSEs of less than 1m over 400m trajectories. Our complete system, called voxgraph, will be available open source.
8 December 2022 in IEEE Robotics and Automation Letters
Integration of multiple sensor modalities and deep learning into Simultaneous Localization And Mapping (SLAM) systems are areas of significant interest in current research. Multi-modality is a stepping stone towards achieving robustness in challenging environments and interoperability of heterogeneous multi-robot systems with varying sensor setups. With maplab 2.0 , we provide a versatile open-source platform that facilitates developing, testing, and integrating new modules and features into a fully-fledged SLAM system. Through extensive experiments, we show that maplab 2.0 ’s accuracy is comparable to the state-of-the-art on the HILTI 2021 benchmark. Additionally, we showcase the flexibility of our system with three use cases: i) large-scale ( $\sim$ 10 $\text{km}$ ) multi-robot multi-session (23 missions) mapping, ii) integration of non-visual landmarks, and iii) incorporating a semantic object-based loop closure module into the mapping framework.