Nathaniel Hanson

I am a Technical Staff Member at the Lincoln Laboratory in the Humanitarian Assistance and Disaster Relief (HADR) Technology group, part of MIT. I work on applications of spectral sensing to environmental monitoring and perception in unstructured environments. My PhD advisors were Taşkın Padır and Kristen Dorsey.

I am always excited to foster new collaborations. If you have any questions about my research, please feel free to contact me.

Email  /  GitHub  /  Google Scholar  /  LinkedIn  /  CV

Biography

Within the HADR group, I lead a portfolio of research focused on the intersection of robotics, disasters, and remote sensing. My PhD research in the Institute for Experiential Robotics focused on making hyperspectral imaging and near-infrared spectroscopy lighter and more cost-effective for use in robot-centric terrain and object understanding. I coined the term "material-informed robotics" to describe a novel reframing of robotics with material perception at its core. I am active in collaborative efforts with MIT campus, including the Beaver Works Summer Institute as the lead instructor for the Unmanned Aerial Vehicle Racing program and the Learning Machines Training course with the MIT Media Lab. I have an MS in Computer Science from Boston University and a BS in Computer Engineering with a minor in Theology from the University of Notre Dame, where I worked in Jane Cleland-Huang's lab.

Research Directions

I'm interested in hyperspectral imaging, material identification, multi-modal sensing, remote sensing, and robotics. A good summary of my research philosophy and interests is provided in my dissertation defense.

Publications

Peer-Reviewed conference and journal papers are listed below. Representative papers are highlighted.

Controlling the Fold: Proprioceptive Feedback in a Soft Origami Robot

We demonstrate proprioceptive feedback control of a one degree of freedom soft, pneumatically actuated origami robot and an assembly of two robots into a two degree of freedom system. Pneumatic actuation, provided by negative fluidic pressure, causes the robot to contract. Capacitive sensors patterned onto the robot provide position estimation and serve as input to a feedback controller. This work contributes optimized capacitive electrode design and the demonstration of closed-loop feedback position control without visual tracking as an input. This approach to capacitance sensing and modeling constitutes a major step towards proprioceptive state estimation and feedback control in soft origami robotics.

PROSPECT: Precision Robot Spectroscopy Exploration and Characterization Tool

Near Infrared (NIR) spectroscopy is vital in industrial quality control for assessing item purity and material quality. We introduce a novel sensorized end effector and acquisition method for capturing spectral signatures and aligning them with a 3D point cloud. Our approach involves a 3D scan of the object, dividing it into planned viewpoints. Robot motion plans ensure optimal spectral signal quality through controlled distance and surface orientation. The system continuously gathers surface reflectance values, creating a four-dimensional model of the object. We demonstrate this system’s effectiveness in profiling complex geometries, showing improved signal consistency compared to naive methods.

HASHI: Highly Adaptable Seafood Handling Instrument for Manipulation in Industrial Settings

We propose a novel robot end effector, called HASHI, that employs chopstick-like appendages for precise and dexterous manipulation. This gripper is capable of in-hand manipulation by rotating its two constituent sticks relative to each other and offers control of objects in all three axes of rotation by imitating human use of chopsticks. HASHI delicately positions and orients food through embedded 6-axis force-torque sensors. We derive and validate the kinematic model for HASHI, as well as demonstrate grip force and torque readings from the sensorization of each chopstick.

Hold 'em and Fold 'em: Towards Human-scale, Feedback-Controlled Soft Origami Robots

In this work, we independently demonstrate these key factors towards controlling and actuating human-scale loads: proprioceptive (embodied) feedback control of a soft, pneumatically-actuated origami robot; and actuation of these origami origami robots under a person’s weight in an open-loop configuration.

Battery-Swapping Multi-Agent System for Sustained Operation of Large Planetary Fleets

This work shares an open-source platform developed to demonstrate battery swapping on unknown field terrain. We detail our design methodologies utilized for increasing system reliability, with a focus on optimization, robust mechanical design, and verification. Optimization of the system is discussed, including the design of passive guide rails through simulation-based optimization methods which increase the valid docking configuration space by 258%. The full system was evaluated during integrated testing, where an average servicing time of 98 seconds was achieved on surfaces with a gradient up to 10 degrees. We conclude by briefly proposing flight considerations for advancing the system toward a space-ready design.

Hyper-Drive: Visible-Short Wave Infrared Hyperspectral Imaging Datasets for Robots in Unstructured Environments

We introduce a first-of-its-kind system architecture with snapshot hyperspectral cameras and point spectrometers to efficiently generate composite datacubes from a moving robot base. Our system collects and registers datacubes spanning the visible to shortwave infrared (660-1700 nm) spectrum while simultaneously capturing the ambient solar spectrum reflected off a white reference tile. We collect and disseminate a large dataset of more than 500 labeled datacubes from on-road and off-road terrain compliant with the ATLAS ontology to further the integration hyperspectral imaging (HSI).

A Vision for Cleaner Rivers: Harnessing Snapshot Hyperspectral Imaging to Detect Macro-Plastic Litter

To address the problem of mismanaged plastic waste, we analyze the feasibility of macro-plastic litter detection using computational imaging approaches in river-like scenarios. We enable near-real-time tracking of partially submerged plastics by using snapshot Visible-Shortwave Infrared hyperspectral imaging. Our experiments indicate that imaging strategies coupled with machine learning classification approaches can lead to high detection accuracy even in challenging scenarios, especially when leveraging hyperspectral data and nonlinear classifiers.

Mobile MoCap: Retroreflector Localization On-The-Go

We present a retroreflector feature detector that performs 6-DoF (six degrees-of-freedom) tracking and operates with minimal camera exposure times to reduce motion blur. To evaluate the proposed localization technique while in motion, we mount our Mobile MoCap system, as well as an RGB camera to benchmark against fiducial markers, onto a precision-controlled linear rail and servo. We evaluate the two systems at varying distances, marker viewing angles, and relative velocities. Across all experimental conditions, our stereo-based Mobile MoCap system obtains higher position and orientation accuracy than the fiducial approach.

SLURP! Spectroscopy of Liquids Using Robot Pre-Touch Sensing

We present a state-of-the-art sensing technique for robots to perceive what liquid is inside of an unknown container. We do so by integrating Visible to Near Infrared (VNIR) reflectance spectroscopy into a robot’s end effector. We introduce a hierarchical model for inferring the material classes of both containers and internal contents given spectral measurements from two integrated spectrometers.

VAST: Visual and Spectral Terrain Classification in Unstructured Multi-Class Environments

To provide mobile robots with the capability to identify the terrain being traversed and avoid undesirable surface types, we propose a multimodal sensor suite capable of classifying different terrains. We capture high resolution macro images of surface texture, spectral reflectance curves, and localization data from a 9 degrees of freedom (DOF) inertial measurement unit (IMU) on 11 different terrains at different times of day. Using this dataset, we train individual neural networks on each of the modalities, and then combine their outputs in a fusion network.

Occluded Object Detection and Exposure in Cluttered Environments with Automated Hyperspectral Anomaly Detection

Our approach proposes a new automated method to perform hyperspectral anomaly detection in cluttered workspaces with the goal of improving robot manipulation. We first assume the dominance of a single material class, and coarsely identify the dominant, non-anomalous class. Our work advances robot perception for cluttered environments by incorporating multi-modal anomaly detection aided by hyperspectral sensing into detecting fractional object presence without need for laboriously curated labels.

Hyperbot-A Benchmarking Testbed For Acquisition Of Robot-Centric Hyperspectral Scene And In-Hand Object Data

We designed a benchmarking testbed to enable a robot manipulator to perceive spectral and spatial characteristics of scene items. Our design includes the use of a push broom Visible to Near Infrared (VNIR) hyperspectral camera, co-aligned with a depth camera. This system enables the robot to process and segment spectral characteristics of items in a larger spatial scene.

In-Hand Object Recognition with Innervated Fiber Optic Spectroscopy for Soft Grippers

We present a novel modular sensing platform integrated into a hybrid-manufactured soft robot gripper to collect and process high-fidelity spectral information. The custom design of the gripper is realized using 3D printing and casting. We embed full-spectrum light sources paired with lensed fiber optic cables within an optically clear gel to collect multi-point spectral reflectivity curves in the Visible to Near Infrared (VNIR) segment of the electromagnetic spectrum.

Other Projects

These include coursework, side projects and unpublished research work.

Material Informed Robotics – Spectral Perception for Object Identification and Parameter Inference

Traditional robot perception has focused on recognizing an object’s semantic purpose as proxy to understanding ideal interaction strategies. However, semantic recognition does not include an implicit understanding of the material composition of these objects. Material recognition aids in understanding properties like weight, friction, and deformability. In this dissertation, the identification of material makeup in objects and scenes composed of heterogeneous materials is achieved through the introduction of near infrared (NIR) spectroscopy into robotics.

Pregrasp Object Material Classification by a Novel Gripper Design with Integrated Spectroscopy

We introduce a novel design for a two fingered gripper with an integrated NIR spectrometer and endoscopic camera to collect VNIR spectral readings and macro surface images from grasped items. We also develop a method based on a nonlinear Support Vector Machine (SVM) to achieve material inference between visually similar items (a smorgasbord of real and artificial fruits) and continually update class estimates with a discrete Bayes filter.

Design and source code from Jon Barron's website