-
Learning Autonomy: Off-Road Navigation Enhanced by Human Input
Authors:
Akhil Nagariya,
Dimitar Filev,
Srikanth Saripalli,
Gaurav Pandey
Abstract:
In the area of autonomous driving, navigating off-road terrains presents a unique set of challenges, from unpredictable surfaces like grass and dirt to unexpected obstacles such as bushes and puddles. In this work, we present a novel learning-based local planner that addresses these challenges by directly capturing human driving nuances from real-world demonstrations using only a monocular camera.…
▽ More
In the area of autonomous driving, navigating off-road terrains presents a unique set of challenges, from unpredictable surfaces like grass and dirt to unexpected obstacles such as bushes and puddles. In this work, we present a novel learning-based local planner that addresses these challenges by directly capturing human driving nuances from real-world demonstrations using only a monocular camera. The key features of our planner are its ability to navigate in challenging off-road environments with various terrain types and its fast learning capabilities. By utilizing minimal human demonstration data (5-10 mins), it quickly learns to navigate in a wide array of off-road conditions. The local planner significantly reduces the real world data required to learn human driving preferences. This allows the planner to apply learned behaviors to real-world scenarios without the need for manual fine-tuning, demonstrating quick adjustment and adaptability in off-road autonomous driving technology.
△ Less
Submitted 14 May, 2025; v1 submitted 25 February, 2025;
originally announced February 2025.
-
myEye2Wheeler: A Two-Wheeler Indian Driver Real-World Eye-Tracking Dataset
Authors:
Bhaiya Vaibhaw Kumar,
Deepti Rawat,
Tanvi Kandalla,
Aarnav Nagariya,
Kavita Vemuri
Abstract:
This paper presents the myEye2Wheeler dataset, a unique resource of real-world gaze behaviour of two-wheeler drivers navigating complex Indian traffic. Most datasets are from four-wheeler drivers on well-planned roads and homogeneous traffic. Our dataset offers a critical lens into the unique visual attention patterns and insights into the decision-making of Indian two-wheeler drivers. The analysi…
▽ More
This paper presents the myEye2Wheeler dataset, a unique resource of real-world gaze behaviour of two-wheeler drivers navigating complex Indian traffic. Most datasets are from four-wheeler drivers on well-planned roads and homogeneous traffic. Our dataset offers a critical lens into the unique visual attention patterns and insights into the decision-making of Indian two-wheeler drivers. The analysis demonstrates that existing saliency models, like TASED-Net, perform less effectively on the myEye-2Wheeler dataset compared to when applied on the European 4-wheeler eye tracking datasets (DR(Eye)VE), highlighting the need for models specifically tailored to the traffic conditions. By introducing the dataset, we not only fill a significant gap in two-wheeler driver behaviour research in India but also emphasise the critical need for developing context-specific saliency models. The larger aim is to improve road safety for two-wheeler users and lane-planning to support a cost-effective mode of transport.
△ Less
Submitted 18 February, 2025;
originally announced February 2025.
-
GO: The Great Outdoors Multimodal Dataset
Authors:
Peng Jiang,
Kasi Viswanath,
Akhil Nagariya,
George Chustz,
Maggie Wigness,
Philip Osteen,
Timothy Overbye,
Christian Ellis,
Long Quang,
Srikanth Saripalli
Abstract:
The Great Outdoors (GO) dataset is a multi-modal annotated data resource aimed at advancing ground robotics research in unstructured environments. This dataset provides the most comprehensive set of data modalities and annotations compared to existing off-road datasets. In total, the GO dataset includes six unique sensor types with high-quality semantic annotations and GPS traces to support tasks…
▽ More
The Great Outdoors (GO) dataset is a multi-modal annotated data resource aimed at advancing ground robotics research in unstructured environments. This dataset provides the most comprehensive set of data modalities and annotations compared to existing off-road datasets. In total, the GO dataset includes six unique sensor types with high-quality semantic annotations and GPS traces to support tasks such as semantic segmentation, object detection, and SLAM. The diverse environmental conditions represented in the dataset present significant real-world challenges that provide opportunities to develop more robust solutions to support the continued advancement of field robotics, autonomous exploration, and perception systems in natural environments. The dataset can be downloaded at: https://www.unmannedlab.org/the-great-outdoors-dataset/
△ Less
Submitted 31 January, 2025;
originally announced January 2025.
-
OTTR: Off-Road Trajectory Tracking using Reinforcement Learning
Authors:
Akhil Nagariya,
Dileep Kalathil,
Srikanth Saripalli
Abstract:
In this work, we present a novel Reinforcement Learning (RL) algorithm for the off-road trajectory tracking problem. Off-road environments involve varying terrain types and elevations, and it is difficult to model the interaction dynamics of specific off-road vehicles with such a diverse and complex environment. Standard RL policies trained on a simulator will fail to operate in such challenging r…
▽ More
In this work, we present a novel Reinforcement Learning (RL) algorithm for the off-road trajectory tracking problem. Off-road environments involve varying terrain types and elevations, and it is difficult to model the interaction dynamics of specific off-road vehicles with such a diverse and complex environment. Standard RL policies trained on a simulator will fail to operate in such challenging real-world settings. Instead of using a naive domain randomization approach, we propose an innovative supervised-learning based approach for overcoming the sim-to-real gap problem. Our approach efficiently exploits the limited real-world data available to adapt the baseline RL policy obtained using a simple kinematics simulator. This avoids the need for modeling the diverse and complex interaction of the vehicle with off-road environments. We evaluate the performance of the proposed algorithm using two different off-road vehicles, Warthog and Moose. Compared to the standard ILQR approach, our proposed approach achieves a 30% and 50% reduction in cross track error in Warthog and Moose, respectively, by utilizing only 30 minutes of real-world driving data.
△ Less
Submitted 5 October, 2021;
originally announced October 2021.
-
An Iterative LQR Controller for Off-Road and On-Road Vehicles using a Neural Network Dynamics Model
Authors:
Akhil Nagariya,
Srikanth Saripalli
Abstract:
In this work we evaluate Iterative Linear Quadratic Regulator(ILQR) for trajectory tracking of two different kinds of wheeled mobile robots namely Warthog (Fig. 1), an off-road holonomic robot with skid-steering and Polaris GEM e6 [1], a non-holonomic six seater vehicle (Fig. 2). We use multilayer neural network to learn the discrete dynamic model of these robots which is used in ILQR controller t…
▽ More
In this work we evaluate Iterative Linear Quadratic Regulator(ILQR) for trajectory tracking of two different kinds of wheeled mobile robots namely Warthog (Fig. 1), an off-road holonomic robot with skid-steering and Polaris GEM e6 [1], a non-holonomic six seater vehicle (Fig. 2). We use multilayer neural network to learn the discrete dynamic model of these robots which is used in ILQR controller to compute the control law. We use model predictive control (MPC) to deal with model imperfections and perform extensive experiments to evaluate the performance of the controller on human driven reference trajectories with vehicle speeds of 3m/s- 4m/s for warthog and 7m/s-10m/s for the Polaris GEM
△ Less
Submitted 28 July, 2020;
originally announced July 2020.
