Aaron Hao Tan

Aaron Hao Tan

University of Toronto

H-index: 4

North America-Canada

About Aaron Hao Tan

Aaron Hao Tan, With an exceptional h-index of 4 and a recent h-index of 4 (since 2020), a distinguished researcher at University of Toronto, specializes in the field of Robotics.

His recent articles reflect a diverse array of research interests and contributions to the field:

4CNet: A Confidence-Aware, Contrastive, Conditional, Consistency Model for Robot Map Prediction in Multi-Robot Environments

NavFormer: A Transformer Architecture for Robot Target-Driven Navigation in Unknown and Dynamic Environments

Deep learning model to identify homonymous defects on automated perimetry

Development of a Pillow Placement Process for Robotic Bed-Making

Deep Reinforcement Learning for Decentralized Multi-Robot Exploration with Macro Actions

Robust Face Mask Detection by a Socially Assistive Robot Using Deep Learning

Design and development of a novel autonomous scaled multiwheeled vehicle

Enhancing Robot Task Completion Through Environment and Task Inference: A Survey from the Mobile Robot Perspective

Aaron Hao Tan Information

University

University of Toronto

Position

___

Citations(all)

83

Citations(since 2020)

83

Cited By

2

hIndex(all)

4

hIndex(since 2020)

4

i10Index(all)

2

i10Index(since 2020)

2

Email

University Profile Page

University of Toronto

Aaron Hao Tan Skills & Research Interests

Robotics

Top articles of Aaron Hao Tan

4CNet: A Confidence-Aware, Contrastive, Conditional, Consistency Model for Robot Map Prediction in Multi-Robot Environments

Authors

Aaron Hao Tan,Siddarth Narasimhan,Goldie Nejat

Journal

arXiv preprint arXiv:2402.17904

Published Date

2024/2/27

Mobile robots in unknown cluttered environments with irregularly shaped obstacles often face sensing, energy, and communication challenges which directly affect their ability to explore these environments. In this paper, we introduce a novel deep learning method, Confidence-Aware Contrastive Conditional Consistency Model (4CNet), for mobile robot map prediction during resource-limited exploration in multi-robot environments. 4CNet uniquely incorporates: 1) a conditional consistency model for map prediction in irregularly shaped unknown regions, 2) a contrastive map-trajectory pretraining framework for a trajectory encoder that extracts spatial information from the trajectories of nearby robots during map prediction, and 3) a confidence network to measure the uncertainty of map prediction for effective exploration under resource constraints. We incorporate 4CNet within our proposed robot exploration with map prediction architecture, 4CNet-E. We then conduct extensive comparison studies with 4CNet-E and state-of-the-art heuristic and learning methods to investigate both map prediction and exploration performance in environments consisting of uneven terrain and irregularly shaped obstacles. Results showed that 4CNet-E obtained statistically significant higher prediction accuracy and area coverage with varying environment sizes, number of robots, energy budgets, and communication limitations. Real-world mobile robot experiments were performed and validated the feasibility and generalizability of 4CNet-E for mobile robot map prediction and exploration.

NavFormer: A Transformer Architecture for Robot Target-Driven Navigation in Unknown and Dynamic Environments

Authors

Haitong Wang,Aaron Hao Tan,Goldie Nejat

Journal

arXiv preprint arXiv:2402.06838

Published Date

2024/2/9

In unknown cluttered and dynamic environments such as disaster scenes, mobile robots need to perform target-driven navigation in order to find people or objects of interest, while being solely guided by images of the targets. In this paper, we introduce NavFormer, a novel end-to-end transformer architecture developed for robot target-driven navigation in unknown and dynamic environments. NavFormer leverages the strengths of both 1) transformers for sequential data processing and 2) self-supervised learning (SSL) for visual representation to reason about spatial layouts and to perform collision-avoidance in dynamic settings. The architecture uniquely combines dual-visual encoders consisting of a static encoder for extracting invariant environment features for spatial reasoning, and a general encoder for dynamic obstacle avoidance. The primary robot navigation task is decomposed into two sub-tasks for training: single robot exploration and multi-robot collision avoidance. We perform cross-task training to enable the transfer of learned skills to the complex primary navigation task without the need for task-specific fine-tuning. Simulated experiments demonstrate that NavFormer can effectively navigate a mobile robot in diverse unknown environments, outperforming existing state-of-the-art methods in terms of success rate and success weighted by (normalized inverse) path length. Furthermore, a comprehensive ablation study is performed to evaluate the impact of the main design choices of the structure and training of NavFormer, further validating their effectiveness in the overall system.

Deep learning model to identify homonymous defects on automated perimetry

Authors

Aaron Hao Tan,Laura Donaldson,Luqmaan Moolla,Austin Pereira,Edward Margolin

Journal

British Journal of Ophthalmology

Published Date

2023/10/1

BackgroundHomonymous visual field (VF) defects are usually an indicator of serious intracranial pathology but may be subtle and difficult to detect. Artificial intelligence (AI) models could play a key role in simplifying the detection of these defects. This study aimed to develop an automated deep learning AI model to accurately identify homonymous VF defects from automated perimetry.MethodsVFs performed on Humphrey field analyser (24–2 algorithm) were collected and run through an in-house optical character recognition program that extracted mean deviation data and prepared it for use in the proposed AI model. The deep learning AI model, Deep Homonymous Classifier, was developed using PyTorch framework and used convolutional neural networks to extract spatial features for binary classification. Total collected dataset underwent 7-fold cross validation for model training and evaluation. To address …

Development of a Pillow Placement Process for Robotic Bed-Making

Authors

Chi-Hong Cheung,Aaron Hao Tan,Andrew Goldenberg

Published Date

2023/8/20

Bed-making is a common chore completed in various living environments to promote user comfort, hygiene, and well-being. Unfortunately, the physical and tedious nature of the act makes it challenging for segments of the elderly community to complete the chore, and thus the opportunity arises to develop robots to automate the task. However, despite the opportunity’s importance and positive impact, there is limited research on developing robotic bed-making systems. The aim of this research is to start addressing this gap by proposing methods for accomplishing pillow placement, a major part of the bed-making task. This paper introduces a pillow placement process to be used by a static 6-DOF (degree of freedom) one-armed robotic manipulator equipped with a 2-finger gripper. The process uses YOLOv4-tiny, image transformations, and principal component analysis (PCA) to infer pillow poses in a …

Deep Reinforcement Learning for Decentralized Multi-Robot Exploration with Macro Actions

Authors

Aaron Hao Tan,Federico Pizarro Bejarano,Yuhan Zhu,Richard Ren,Goldie Nejat

Journal

IEEE Robotics and Automation Letters

Published Date

2023/1/1

Cooperative multi-robot teams need to be able to explore cluttered and unstructured environments while dealing with communication dropouts that prevent them from exchanging local information to maintain team coordination. Therefore, robots need to consider high-level teammate intentions during action selection. In this letter, we present the first Macro Action Decentralized Exploration Network (MADE-Net) using multi-agent deep reinforcement learning (DRL) to address the challenges of communication dropouts during multi-robot exploration in unseen, unstructured, and cluttered environments. Simulated robot team exploration experiments were conducted and compared against classical and DRL methods where MADE-Net outperformed all benchmark methods in terms of computation time, total travel distance, number of local interactions between robots, and exploration rate across various degrees of …

Robust Face Mask Detection by a Socially Assistive Robot Using Deep Learning

Authors

Yuan Zhang,Meysam Effati,Aaron Hao Tan,Goldie Nejat

Journal

Computers

Published Date

2023/12/23

Wearing masks in indoor and outdoor public places has been mandatory in a number of countries during the COVID-19 pandemic. Correctly wearing a face mask can reduce the transmission of the virus through respiratory droplets. In this paper, a novel two-step deep learning (DL) method based on our extended ResNet-50 is presented. It can detect and classify whether face masks are missing, are worn correctly or incorrectly, or the face is covered by other means (e.g., a hand or hair). Our DL method utilizes transfer learning with pretrained ResNet-50 weights to reduce training time and increase detection accuracy. Training and validation are achieved using the MaskedFace-Net, MAsked FAces (MAFA), and CelebA datasets. The trained model has been incorporated onto a socially assistive robot for robust and autonomous detection by a robot using lower-resolution images from the onboard camera. The results show a classification accuracy of 84.13% for the classification of no mask, correctly masked, and incorrectly masked faces in various real-world poses and occlusion scenarios using the robot.

Design and development of a novel autonomous scaled multiwheeled vehicle

Authors

Aaron Hao Tan,Michael Peiris,Moustafa El-Gindy,Haoxiang Lang

Journal

Robotica

Published Date

2022/5

This article proposes the design and development of a novel custom-built, autonomous scaled multiwheeled vehicle that features an eight-wheel drive and eight-wheel steer system. In addition to the mechanical and electrical design, high-level path planning and low-level vehicle control algorithms are developed and implemented including a two-stage autonomous parking algorithm is developed. A modified position-based visual servoing algorithm is proposed and developed to achieve precise pose correction. The results show significant gains in accuracy and efficiency comparing with an open-source path planner. It is the aim of this work to expand the research of autonomous platforms taking the form of commercial and off-road vehicles using actuated steering and other mechanisms attributed to passenger vehicles. The outcome of this work is a unique autonomous research platform that features …

Enhancing Robot Task Completion Through Environment and Task Inference: A Survey from the Mobile Robot Perspective

Authors

Aaron Hao Tan,Goldie Nejat

Published Date

2022/12

In real-world environments, ranging from urban disastrous scenes to underground mining tunnels, autonomous mobile robots are being deployed in harsh and cluttered environments, having to deal with perception and communication issues that limit their facilitation for data sharing and coordination with other robots. In these scenarios, mobile robot inference can be used to increase spatial awareness and aid decision-making in order to complete tasks such as navigation, exploration, and mapping. This is advantageous as inference enables robots to plan with predicted information that is otherwise unobservable, thus, reducing the replanning efforts of robots by anticipating future states of both the environment and teammates during execution. While detailed reviews have explored the use of inference during human–robot interactions, to-date none have explored mobile robot inference in unknown environments …

A Sim-to-Real Pipeline for Deep Reinforcement Learning for Autonomous Robot Navigation in Cluttered Rough Terrain

Authors

Han Hu,Kaicheng Zhang,Aaron Hao Tan,Michael Ruan,Christopher Agia,Goldie Nejat

Journal

IEEE Robotics and Automation Letters

Published Date

2021/6/30

Robots that autonomously navigate real-world 3D cluttered environments need to safely traverse terrain with abrupt changes in surface normals and elevations. In this letter, we present the development of a novel sim-to-real pipeline for a mobile robot to effectively learn how to navigate real-world 3D rough terrain environments. The pipeline uses a deep reinforcement learning architecture to learn a navigation policy from training data obtained from the simulated environment and a unique combination of strategies to directly address the reality gap for such environments. Experiments in the real-world 3D cluttered environment verified that the robot successfully performed point-to-point navigation from arbitrary start and goal locations while traversing rough terrain. A comparison study between our DRL method, classical, and deep learning-based approaches showed that our method performed better in terms of …

See List of Professors in Aaron Hao Tan University(University of Toronto)

Aaron Hao Tan FAQs

What is Aaron Hao Tan's h-index at University of Toronto?

The h-index of Aaron Hao Tan has been 4 since 2020 and 4 in total.

What are Aaron Hao Tan's top articles?

The articles with the titles of

4CNet: A Confidence-Aware, Contrastive, Conditional, Consistency Model for Robot Map Prediction in Multi-Robot Environments

NavFormer: A Transformer Architecture for Robot Target-Driven Navigation in Unknown and Dynamic Environments

Deep learning model to identify homonymous defects on automated perimetry

Development of a Pillow Placement Process for Robotic Bed-Making

Deep Reinforcement Learning for Decentralized Multi-Robot Exploration with Macro Actions

Robust Face Mask Detection by a Socially Assistive Robot Using Deep Learning

Design and development of a novel autonomous scaled multiwheeled vehicle

Enhancing Robot Task Completion Through Environment and Task Inference: A Survey from the Mobile Robot Perspective

...

are the top articles of Aaron Hao Tan at University of Toronto.

What are Aaron Hao Tan's research interests?

The research interests of Aaron Hao Tan are: Robotics

What is Aaron Hao Tan's total number of citations?

Aaron Hao Tan has 83 citations in total.

    academic-engine

    Useful Links