About

Junaed Sattar is an associate professor in the Department of Computer Science & Engineering at the University of Minnesota Twin Cities. His primary research focus is on making robots work safely and intuitively with people, enabling humans and robots to coexist and collaborate effectively. This involves improving a robot's perception of people, their intentions, and actions, as well as engaging in dialog and environmental understanding. His research addresses the challenge of perceiving the world robustly under changing and degraded conditions, extending into multi-modal sensory perception. Sattar's past research has been heavily influenced by developing robots that operate in unstructured environments, particularly underwater, and his current work involves field robots in air, water, and outdoor all-terrain platforms. He is the founding director of the Minnesota Interactive Robotics and Vision Laboratory and is a member of the IEEE. Sattar joined the department in 2016 as an assistant professor and was promoted to associate professor in 2022. His educational background includes a Ph.D. and M.S. in Computer Science/Robotics from McGill University and a B.S. in Computer Science and Engineering from Bangladesh University of Engineering and Technology. His professional experience includes postdoctoral work at the University of British Columbia and a faculty position at Clarkson University before joining the University of Minnesota.

Research topics

• Computer science
• Artificial intelligence
• Computer vision
• Human–computer interaction
• Real-time computing

Selected publications

• AMP2026: A Multi-Platform Marine Robotics Dataset for Tracking and Mapping

  2026-05-14

  articleOpen access

  Marine environments present significant challenges for perception and autonomy due to dynamic surfaces, limited visibility, and complex interactions between aerial, surface, and submerged sensing modalities. This paper introduces the Aerial–Marine Perception Dataset (AMP2026), a multi-platform marine robotics dataset collected across multiple field deployments designed to support research in two primary areas: multi-view tracking and marine environment mapping. The dataset includes synchronized data from aerial drones, boat-mounted cameras, and submerged robotic platforms, along with associated localization and telemetry information. The goal of this work is to provide a publicly available dataset enabling research in marine perception and multi-robot observation scenarios. This paper describes the data collection methodology, sensor configurations, dataset organization, and intended research tasks supported by the dataset.

  PublisherDOI

• Robotic estimation of single scuba diver respiration rate for safety in underwater human-robot collaboration

  The International Journal of Robotics Research · 2026-01-08

  articleOpen accessSenior authorCorresponding

  Human respiration rate (HRR) is an important physiological metric for diagnosing a variety of health conditions from stress levels to heart conditions. Estimation of HRR is well-studied in controlled terrestrial environments, yet robotic estimation of HRR as an indicator of scuba diver stress for underwater human-robot interaction (UHRI) scenarios is hitherto unexplored. In this paper, we introduce a novel approach for robotic estimation of HRR from underwater visual data by observing the volume of bubbles from exhalation states in scuba diving to measure the respiration rate. We propose and exploit a fuzzy labeling system that uses audio information to annotate exhalation and inhalation states based on the presence of bubbles. We collect an expansive audio and visual dataset of diver breathing from diverse geographical locations and water bodies, on which we compare four different methods for classifying the presence of bubbles in images. We then use a novel estimation algorithm to assess the diver respiration state from visual classification for computing the HRR in breaths per minute. Ultimately, we demonstrate the efficacy of our method in estimating HRR by comparing the respiration rate output against those measured by human analysts.

  PublisherDOI

• Diver interest via pointing in three dimensions: 3D pointing reconstruction for diver-AUV communication

  Autonomous Robots · 2026-04-20

  preprintOpen accessSenior author

  Abstract This paper presents Diver Interest via Pointing in Three Dimensions (DIP-3D), a method to indicate an object of interest from a diver to an autonomous underwater vehicle (AUV) by pointing that includes three-dimensional distance information to discriminate between multiple objects in the AUV’s field of view. Traditional dense stereo vision for distance estimation underwater is challenging because of the relative lack of saliency of scene features and degraded lighting conditions. Yet in many applications, including distance information is necessary for robotic perception of diver pointing when multiple objects appear within the robot’s image view. We subvert the challenges of underwater distance estimation by using sparse reconstruction of specific keypoints in both the left and right images from the robot’s stereo camera to perform pose estimation. Triangulated pose keypoints, along with any object detection method, enable DIP-3D to infer the location of an object of interest when multiple objects are in the AUV’s field of view. By allowing the scuba diver to point at an arbitrary object of interest and enabling the AUV to autonomously decide which object the diver is pointing to, this method permits more natural interaction between AUVs and humans in underwater-human robot collaborative tasks.

  PublisherOA PDFDOI

• AMP2026: A Multi-Platform Marine Robotics Dataset for Tracking and Mapping

  ArXiv.org · 2026-03-04

  articleOpen access

  Marine environments present significant challenges for perception and autonomy due to dynamic surfaces, limited visibility, and complex interactions between aerial, surface, and submerged sensing modalities. This paper introduces the Aerial Marine Perception Dataset (AMP2026), a multi-platform marine robotics dataset collected across multiple field deployments designed to support research in two primary areas: multi-view tracking and marine environment mapping. The dataset includes synchronized data from aerial drones, boat-mounted cameras, and submerged robotic platforms, along with associated localization and telemetry information. The goal of this work is to provide a publicly available dataset enabling research in marine perception and multi-robot observation scenarios. This paper describes the data collection methodology, sensor configurations, dataset organization, and intended research tasks supported by the dataset.

  PublisherOA PDF

• AMP2026: A Multi-Platform Marine Robotics Dataset for Tracking and Mapping

  Open MIND · 2026-03-04

  preprint

  Marine environments present significant challenges for perception and autonomy due to dynamic surfaces, limited visibility, and complex interactions between aerial, surface, and submerged sensing modalities. This paper introduces the Aerial Marine Perception Dataset (AMP2026), a multi-platform marine robotics dataset collected across multiple field deployments designed to support research in two primary areas: multi-view tracking and marine environment mapping. The dataset includes synchronized data from aerial drones, boat-mounted cameras, and submerged robotic platforms, along with associated localization and telemetry information. The goal of this work is to provide a publicly available dataset enabling research in marine perception and multi-robot observation scenarios. This paper describes the data collection methodology, sensor configurations, dataset organization, and intended research tasks supported by the dataset.

  DOI

• Robotic Detection of Deformed Objects for Marine Debris Management

  2025-09-29

  articleSenior author

  The growing presence of trash in our water bodies highlights the urgency of removing marine debris. Autonomous underwater vehicles (AUVs) equipped with object detection capabilities can play a vital role in the removal of marine trash. While significant improvements have been made for terrestrial object detection, detection in the underwater domain suffers from environmental challenges and data scarcity. Additionally, objects erode and deform when submerged underwater, limiting the effectiveness of existing datasets. In this work, we explore various detection methods and their applicability in the underwater domain. We evaluate two types of methods: feature extractor-based methods and data generation methods. We attempt to identify deformation-invariant features using existing feature extractors and detection networks. We also generate relevant deformation data using style transfer and diffusion-based methods. We find that style transfer-based blending performs better than diffusion-based methods for image generation. After evaluating the methods, we highlight the need for relevant training datasets and the need to explore networks that can account for deformations in objects.

  PublisherDOI

• Design and Development of the MeCO Open-Source Autonomous Underwater Vehicle

  ArXiv.org · 2025-03-13

  preprintOpen accessSenior author

  We present MeCO, the Medium Cost Open-source autonomous underwater vehicle (AUV), a versatile autonomous vehicle designed to support research and development in underwater human-robot interaction (UHRI) and marine robotics in general. An inexpensive platform to build compared to similarly-capable AUVs, the MeCO design and software are released under open-source licenses, making it a cost effective, extensible, and open platform. It is equipped with UHRI-focused systems, such as front and side facing displays, light-based communication devices, a transducer for acoustic interaction, and stereo vision, in addition to typical AUV sensing and actuation components. Additionally, MeCO is capable of real-time deep learning inference using the latest edge computing devices, while maintaining low-latency, closed-loop control through high-performance microcontrollers. MeCO is designed from the ground up for modularity in internal electronics, external payloads, and software architecture, exploiting open-source robotics and containerarization tools. We demonstrate the diverse capabilities of MeCO through simulated, closed-water, and open-water experiments. All resources necessary to build and run MeCO, including software and hardware design, have been made publicly available.

  PublisherOA PDFDOI

• Reactions and Engagement of Individuals with Dementia Toward Humanoid Assistive Robots: A Study Using the Pepper Robot

  Studies in health technology and informatics · 2025-05-12

  articleOpen access

  While robot acceptance in different populations is well-studied, little is known about how individuals with dementia perceive and respond to humanoid assistive robots. This paper explores how individuals affected by dementia react to and engage with such robots, focusing on interactions with Pepper, a humanoid robot. Conducted in an all-dementia nursing home with residents experiencing varying stages of dementia, the study has collected direct observations and participant feedback. A common concern among clinicians, family members, and caregivers is that individuals with dementia may find robots frightening or unsettling, raising questions about their suitability for caregiving roles. However, the findings of this study suggest otherwise. Residents consistently identified the robots as "cute" and "child-like," with many expressing comfort and interest in interacting with them. These results highlight the potential for humanoid robots like Pepper to serve as non-threatening, engaging companions for individuals with dementia, addressing caregiving needs while enhancing their well-being. This study provides a foundation for further exploration into the acceptance and application of assistive robotics in dementia care settings.

  PublisherOA PDFDOI

• Stereo-Based 3D Human Pose Estimation for Underwater Robots Without 3D Supervision

  IEEE Robotics and Automation Letters · 2025-04-03 · 4 citations

  articleSenior author

  In this paper, we propose a novel deep learning-based 3D underwater human pose estimator capable of providing metric 3D poses of scuba divers from stereo image pairs. While existing research has made significant advancements in 3D human pose estimation, most methods rely on 3D ground truth for training, which is challenging to acquire in dynamic environments where traditional motion capture systems are impractical to deploy. To overcome this, our approach leverages epipolar geometry to derive 3D information from 2D estimations. Our method estimates semantic keypoints while capturing their corresponding disparity from binocular perspectives, thus avoiding challenges in calibrating for multi-view setups or scale-ambiguity in monocular settings. Additionally, to reduce the sensitivity of our method to 2D annotation accuracy, we propose an auto-refinement pipeline to automatically correct biases introduced by human labeling. Experiments demonstrate that our approach significantly improves performance compared to previous state-of-the-art methods in different environments, including but not limited to underwater scenarios, while being efficient enough to run on limited-capacity edge devices.

  PublisherDOI

• A Quantitative Evaluation of Bathymetry-Based Bayesian Localization Methods for Autonomous Underwater Robots

  IEEE Journal of Oceanic Engineering · 2025-03-24

  articleSenior author

  This article presents an evaluation of four probabilistic algorithms for bathymetry-based localization of autonomous underwater vehicles (AUVs). The algorithms fuse a priori bathymetry information with depth and range measurements to localize an AUV underwater using four different Bayes filters [extended Kalman filter, unscented Kalman filter, particle filter, and marginalized PF (MPF)]. We develop the algorithms using the robot operating system (ROS), build a realistic simulation platform using ROS Gazebo incorporating real-world bathymetry, and evaluate the performance of these four Bayesian bathymetry-based AUV localization approaches on real-world lake data. The simulation allows the evaluation of algorithms with accurate knowledge of the robot's true location, which is otherwise infeasible to obtain underwater in the field. By relying on the data from a depth sensor and echo sounder, the localization algorithms overcome challenges faced by visual landmark-based localization. Our results show the efficacy of each algorithm under a variety of conditions, with the MPF being the most accurate in general.

  PublisherDOI

Recent grants

• NRI: Collaborative Research: Autonomous Quadrotors for 3D Modeling and Inspection of Outdoor Infrastructure

  NSF · $830k · 2016–2022

• Towards Robust and Natural Underwater Human-Robot Interaction

  NSF · $142k · 2019–2021

Frequent coauthors

• Gregory Dudek

  39 shared

• Michael Fulton

  University of Minnesota

  38 shared

• Jiawei Mo

  28 shared

• Md Jahidul Islam

  27 shared

• Jungseok Hong

  23 shared

• Chelsey Edge

  21 shared

• Peigen Luo

  Twin Cities Orthopedics

  21 shared

• Jahidul Islam

  20 shared

Labs

• Minnesota Interactive Robotics and Vision LaboratoryPI

Awards & honors

• Selected Grants NRI: Enhancing Autonomous Underwater Robot P…
• NRI: Collaborative Research: Autonomous Quadrotors for 3D Mo…
• EAGER: Towards robust and natural underwater human-robot int…