BEGIN:VCALENDAR CALSCALE:GREGORIAN VERSION:2.0 METHOD:PUBLISH PRODID:-//Drupal iCal API//EN X-WR-TIMEZONE:America/New_York BEGIN:VTIMEZONE TZID:America/New_York BEGIN:DAYLIGHT TZOFFSETFROM:-0500 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU DTSTART:20070311T020000 TZNAME:EDT TZOFFSETTO:-0400 END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:-0400 RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU DTSTART:20071104T020000 TZNAME:EST TZOFFSETTO:-0500 END:STANDARD END:VTIMEZONE BEGIN:VEVENT SEQUENCE:1 X-APPLE-TRAVEL-ADVISORY-BEHAVIOR:AUTOMATIC UID:235731 DTSTAMP:20260428T092652Z DTSTART;TZID=America/New_York:20260501T173000 DTEND;TZID=America/New_York:20260501T190000 URL;TYPE=URI:https://www.wpi.edu/news/calendar/events/robotics-engineering- spring-2026-graduate-capstone SUMMARY:Robotics Engineering Spring 2026 Graduate Capstone DESCRIPTION:\n\n\n \n \n\n\n\nRobotics Engineering Spring 2026 Gr aduate Capstone Schedule\n\n\n5:30 PM - 6:00 PM\nSIMRAN CHAUHAN\nTitle: Re al-Time Direction-Aware Traversability Estimation with Physics Supervision for Off-Road Autonomous Vehicles\nAbstract: Off-road autonomous vehicles require accurate, real-time assessment of terrain traversability that acco unts for terrain topography, and vehicle geometry and orientation. However , performing this assessment through conventional physics-based mobility a nalysis is computationally prohibitive for real-time use. This work presen ts a learning-based approach by incorporating a novel criteria for topogra phical mobility evaluation and thus provides a fast alternative to computa tionally expensive physics-based mobility analysis. The proposed method is trained using supervision from a deterministic mobility engine and is des igned to identify safety-critical impassable regions under significant cla ss imbalance. By incorporating terrain-geometry information during learnin g, the approach captures richer mobility cues than binary passability labe ls alone. The framework is evaluated on synthetic off-road terrains of var ying difficulty and demonstrated on simulated LiDAR-derived terrain maps f rom simulation. Results show improved detection of impassable terrain and produce aggregated traversability scores that can be directly used for dow nstream path planning, supporting real-time mobility assessment for autono mous ground vehicles in unstructured environments.\nAdvisor: Lee Moradi\nP AVAN GANESH PABBINEEDI\nTitle: On the Advantages of Control-Space Kinodyna mic Planning and Control Over Geometric State-Space Methods for Ground Veh icle Navigation\nAbstract: Autonomous ground vehicles need motion planning and control methods that consider the vehicle's dynamic limitations. Trad itional approaches first plan a path in the geometric state space using on ly position and heading, and then use a controller to follow that path bas ed mainly on positional error. Because these methods ignore velocity and d ynamic feasibility during planning, the controller must make reactive corr ections to follow trajectories that may not be dynamically achievable, lea ding to larger tracking errors and less smooth motion. To address this, th is project implements a motion planning and control framework that operate s directly in the control space. It generates time-parameterized reference s that are dynamically feasible by construction. The controller consumes t his full kinodynamic reference, enabling predictive control that accounts for vehicle kinodynamics, rather than tracking geometric waypoints alone.T he framework is tested on two platforms with different dynamics: an Ackerm ann-steered vehicle in simulation and a differential-drive robot in a real indoor environment. The proposed approach is compared with a traditional geometric state-space method using the same controller and optimizer. Resu lts show a significant reduction in cross-track error, with the kinodynami c control-space approach achieving lower tracking error while also produci ng smoother motion. These results demonstrate the benefits of integrating planning and control directly in the control space for autonomous vehicle navigation.\nAdvisor: Constantinos Chamzas\nMANIDEEP DUGGI\nTitle: A Prefe rence-Aware, Socially Aware Multi-Robot Assistive System\nAbstract: This p roject presents a socially and physically assistive multi-robot system tha t integrates a social robot, Pepper, with a mobile manipulation robot, Str etch, to support object-fetch and delivery tasks during daily activities. In this system, Stretch provides physical assistance through autonomous na vigation, object retrieval, and delivery, while Pepper serves as the prima ry user-facing robot, providing spoken interaction, social engagement, tas k explanations, confirmation prompts, and progress updates. To make the in teraction more personalized and user-centered, the project investigates in dividual differences in preferences for robot communication and assistance style. Based on user responses to a structured preference questionnaire, the system defines multiple interaction profiles that vary along dimension s such as verbal presence, update frequency, initiative, consent, social t one, explanation detail, support, and anthropomorphic style. These profile s are implemented as distinct robot characters that allow Pepper to tailor its behavior to different user preferences. Overall, the project demonstr ates how a coordinated multi-robot assistive system can combine physical t ask support with personalized social interaction to improve user comfort, engagement, transparency, and acceptance during everyday assistive tasks, with the potential to promote long-term user adoption of assistive robots. \nAdvisor: Fiona Yuan\nROBERT GUNDUZ\nTitle: Reinforcement Learning for Fr ontier Exploration\nAbstract: Exploring new reinforcement learning methodo logy for tasking a robot to map an unknown environment as quickly as possi ble. Implementation includes, but is not limited to, tuning hyper-paramete rs for an existing Advantage Actor Critic (A2C) model, or investigating ot her approaches such as sequence to sequence (Sec2Sec).\nAdvisor: Kevin Lea hy\nDHRUV AGRAWAL\nTitle: EAGLE: Efficient Aerial Guidance for Interconnec ted Load Systems\nAbstract: This work presents a control framework for coo perative aerial payload manipulation using single and multiple quadrotors transporting a cable-suspended load. The system considers both a quadrotor -centric formulation, where UAVs independently track trajectories while tr eating the payload as a disturbance, and a payload-centric formulation, wh ere the coupled dynamics of the quadrotor–payload system are explicitly mo deled for direct payload trajectory control. A geometric controller is emp loyed to ensure stable and precise flight, while a multi-quadrotor extensi on enables cooperative transport of a rigid body payload through taut cabl e connections. The proposed framework enables coordinated aerial manipulat ion by leveraging system-level dynamics to achieve stable and accurate pay load trajectory tracking under coupling constraints.\nAdvisor: Guanrui Li\ nLUCAS BURSTEIN\nTitle: Robotically Controlled Magnetic Laser Aiming for S urgical Applications\nAbstract: This project presents a novel optical fibe r steering system for laser surgery based on magnetic actuation. Fiber ste ering is achieved by means of two permanent magnets: a small magnet mounte d at the distal end of the fiber and a larger, robotically manipulated mag net that generates controlled magnetic torque and force. In clinical use, the larger magnet would remain outside the patient’s body, positioned clos e enough to produce the necessary magnetic fields. Open-loop laser aiming of the proposed system was demonstrated.\nAdvisor: Loris Fichera\n6:00 PM - 6:30 PM\nWARWICK BARKER\nTitle: Facilitating Industrial Changeover Throu gh A Sketch-Based Interface\nAbstract: Industrial changeover is a wasteful process that continues to hinder automation efforts in high mix, low volu me (HMLV) manufacturing environments. Without a way to mitigate changeover waste, it is difficult to economically justify new automation projects in HMLV industries. This project aims to reduce wasteful changeover time for industrial robot applications through the use of a sketch-based interface and robot-mounted camera. With little-to-no training, an operator can set up a new job using simple sketching tools to draw on an image of the new part. This process further abstracts programming from changeover operation s to the point that an operator needs zero understanding of conventional t each pendant programming methods to successfully complete a changeover ope ration. By simplifying changeover and reducing waste, manufacturers can ju stify and implement increased automation in one of the most automation-ave rse manufacturing segments.\nAdvisor: Jane Li\nSEYEDALI GOLESTANEH\nTitle: AURA: Asymptotically Optimal Uncertainty-Robust Replanning Algorithm for Kinodynamic Planning\nAbstract: Sampling-based motion planners offer a pra ctical and scalable approach to kinodynamic motion planning, notably for h igh-dimensional, underactuated, or non-holonomic systems, where analytic t rajectory generation and exhaustive search over the state space are infeas ible. However, their sequential computation leads to slow convergence towa rd optimal solutions. Moreover, the offline planning process cannot be cou pled with execution steps to take into account of the motion uncertainty. In this work, both limitations are addressed within a unified framework, A URA, an asymptotically optimal meta-planner, robust to uncertainties that arise during execution. Alongside the main execution thread, this framewor k consists of a replanning module, which continuously explores the state s pace and improves the plan during execution, and an optimization module, w hich refines control inputs to reduce tracking error. The proposed approac h is evaluated in both simulation and real-world environments across a var iety of systems, demonstrating consistent improvements in path efficiency in comparison to other baselines.\nAdvisor: Constantinos Chamzas\nTANMAYI INAPARTHY\nTitle: Beyond Physical Safety: Vision-Language Models for Emoti onal Safety Reasoning in Dementia Care\nAbstract: SafeScore is a framework that fine-tunes vision-language models to reason about emotional safety i n dementia care interactions. Rather than focusing solely on physical haza rds, SafeScore analyzes care videos to interpret the emotional states of p eople living with dementia (PLWDs) — processing facial expressions, body l anguage, gestures, and speech together — and generates structured safety s cores that quantify whether a caregiver's actions are helping or unintenti onally escalating distress. The system builds a multimodal training pipeli ne using dementia care videos, NVILA-8B-generated captions, Whisper transc ripts, and DeepSeek-R1-generated Question–Reasoning–Answer supervision to fine-tune long-context VLMs for emotional reasoning. In addition to the la rge-model pipeline, We also aim to develop a lightweight SafeScore using M oondream2, a lightweight VLM, enabling emotional safety scoring to run eff iciently on small, resource-constrained devices — making the framework dep loyable in real-world assistive robots, smart-home systems, and telehealth platforms where computational resources may be limited.\nAdvisor: Fiona Y uan\nHARMEET DHILLON\nTitle: Real Time optical Communication using event c ameras\nAbstract: In multi-robot systems, traditional radio frequency (RF) communication struggles with contention and jamming. Optical communicatio n offers a strong alternative. However, conventional frame-based cameras s uffer from limited frame rates, motion blur, and reduced robustness under high dynamic range lighting. Event cameras support microsecond temporal re solution and high dynamic range, making them extremely sensitive to scene changes under fast relative motion with an optical transmitter. Leveraging these strengths, we develop a complete optical communication system capab le of tracking moving transmitters and decoding messages in real time. Our system achieves over 95% decoding accuracy for text transmission during m otion by implementing a Geometry Aware Unscented Kalman Filter (GA-UKF), a chieving 7× faster processing speed compared to the previous state of-the- art method, while maintaining equivalent tracking accuracy at transmitting frequencies ≥ 1 kHz.\nAdvisor: Kevin Leahy\nYASAR IDIKUT\nTitle: Safety-C ritical Quadrotor Navigation Using Fourth-Order Control Barrier Functions with Full Nonlinear Quadrotor Dynamics\nAbstract: Safety-critical quadroto r navigation in cluttered environments typically relies on hierarchical ar chitectures in which a high-level safety filter modifies position referenc es that are then tracked by a separate attitude controller. However, This separation can create a mismatch between translational safety objectives a nd attitude evolution. To address this gap, we propose a safety controller based on high order control barrier functions that explicitly incorporate s the full nonlinear quadrotor dynamics, including attitude dynamics and t orque-level actuation. Moreover, we propose 2 control designs to leverage the nonlinear model predictive control and the proposed high-order control barrier function for the tasks of quadrotor obstacle avoidance. The resul ting formulation enforces obstacle-avoidance constraints while optimizing control torques, granting the safety layer direct authority over attitude behavior and eliminating the hierarchical disconnect inherent to reference -tracking safety filters. We validate the approaches in both high-fidelity simulation and real-world flight experiments, demonstrating reliable cons traint satisfaction and successful obstacle avoidance. These results sugge st that torque-level high-order CBFs provide a practical pathway to provab ly safer quadrotor navigation under full-body dynamics.\nAdvisor: Guanrui Li\nJESSICA HART\nTitle: SWAMP Docking Station Design\nAbstract: The Syste m for Watering and Autonomously Monitoring Plants (SWAMP) is a mobile robo t designed to care for plants with minimal user input. During its developm ent by last year's MQP team, the robot successfully demonstrated navigatio n, obstacle avoidance, and precision watering. However, its autonomy was l imited by the need for manual water refilling and battery monitoring. This capstone addresses those limitations through the design and development o f a docking station for SWAMP. The docking station will serve as a home ba se and autonomously refill the robot's onboard water tanks. The addition o f a docking station will reduce required user maintenance and enable SWAMP to care for plants over extended periods without human intervention.\nAdv isor: Greg Lewin\n6:30 PM - 7:00 PM\nEVAN CARMODY\nTitle: Gripper Design f or Soft Continuum Robots\nAbstract: Designing and fabricating a lightweigh t, effective gripper design for implementation with a cable-driven origami module soft continuum robot arm. The gripper will contain a camera for vi sion based servoing and must integrate well with control and actuation pip eline. Testing and integration with the entire system will follow to evalu ate designs for efficiency and compatibility.\nAdvisor: Berk Calli\nJOHN H ALL\nTitle: Structured Light for Endoscopic Depth Estimation\nAbstract: Cu rrent gastric endoscopy vision capabilities are limited by the use of mono cular cameras. Estimating the volume of the stomach is a crucial component for successful Endoscopic Sleeve Gastroplasty (ESG) operations and is cur rently done by surgeon intuition. Traditional vision based depth estimatio ns such as stereo or time of flight struggle in the poorly lit and reflect ive environment of the stomach. The solution proposed by this project is d epth estimation using structured light with a monocular camera and diffrac ted laser system. The system generates a sparse but robust point cloud whi ch can be used to estimate the volume of the stomach. This project outline s a method to calibrate and camera/laser pair and estimate the depth. The system was tested on a variety of surface shapes and materials in addition to ex-vivo stomach tissue.\nAdvisor: Giovanni Pittiglio\nAMEYA PHADNIS\nT itle: An Improved Manipulation Framework for Object Grasping using Point C louds\nAbstract: This Capstone Project proposes a consolidated vision-guid ed grasping pipeline that leverages a novel octree-based vision algorithm applied to point clouds that is designed to provide a compact yet geometri cally informative description of the scene. Building on this representatio n, a new algorithmic grasping method is developed that directly operates o n the octree structure, enabling efficient reasoning over object geometry and spatial occupancy without reliance on learning-based models. The propo sed framework is tested in Gazebo on several objects from the YCB dataset, and the success rate and grasp pose accuracy for the objects are noted. I t is observed that with the help of the vision output, a simple approach i s required to perform grasping of objects, which would otherwise require h eavy computation or learning-based methods to achieve similar results.\nAd visor: Jing Xiao\nBRIJAN VAGHASIYA\nTitle: Navigation in Complete Darkness using Neuromorphic Cameras\nAbstract: This research presents a method for autonomous navigation in zero-light environments using Neuromorphic camer a, structured lighting and coded apertures. By projecting structured light patterns and capturing the reflected images through a coded aperture, the system enhances depth estimation in complete darkness. A deep learning fr amework processes these images to extract depth information, enabling navi gation in challenging, low-visibility environments.\nAdvisor: Nitin Sanket \n END:VEVENT END:VCALENDAR