Emad Razavi
Robotics Engineer (MSc) focused on perception and navigation research for mobile robots, with a background in electronics, embedded systems, and control theory, and hands-on robotics system development experience.
Robotics Engineer (MSc) focused on perception and navigation research for mobile robots, with a background in electronics, embedded systems, and control theory, and hands-on robotics system development experience.
My path into robotics started from electrical engineering, where I built a foundation in electronics, embedded systems, control theory, and optimization. That background shaped the way I approach robots: not as isolated algorithms, but as complete systems where sensing, computation, actuation, and reliability have to work together.
During my robotics studies, I moved deeper into perception and navigation for mobile robots, especially systems that can leave simulation and operate on real platforms. My work has included semantic mapping, object-goal navigation, ROS 2 integration, Nav2-based autonomy, and experiments with physical robots.
I am especially interested in research and engineering that connects perception with action: robots that understand enough about their environment to move safely, make useful decisions, and support people in real-world tasks.
My academic background combines robotics engineering, electrical engineering, embedded systems, control theory, and optimization.
University of Genoa, 2023-2025.
Thesis: Semantic Object-Goal Navigation on a Quadruped Robot in Known Environments
Open master thesis explanation
Qazvin Azad University, 2018-2023.
Thesis: Optimal PID Controller Tuning for Time-Delayed Systems Using Ant Colony Optimization
Open bachelor thesis explanation
Semantic Object-Goal Navigation on a Quadruped Robot in Known Environments
This thesis developed a two-stage object-goal navigation pipeline for Boston Dynamics Spot. The system first records a clean 2D map and a compact set of confirmed object instances, then reuses that saved world model to navigate to named objects safely and repeatably.
A short teleoperation run builds the occupancy map and records only stable RGB D object observations into a semantic database.
Spot localizes on the saved map, selects a requested object, and sends Nav2 a safe standoff goal near that object. Targets can be selected from a CLI or by voice.
Work carried out at Dynamic Legged Systems, Istituto Italiano di Tecnologia.
The stack is intentionally lightweight: Spot uses a 2D LiDAR for mapping/localization, a RealSense T265 for visual-inertial odometry, and a RealSense D435 for RGB D detections. SLAM Toolbox builds the pre-run map; during navigation, AMCL localizes on the fixed map and Nav2 sends goals through the ROS 2 bridge.
During the pre run, D435 RGB detections are projected into 3D using depth and transformed into the map frame. The semantic layer keeps temporary observations separate from confirmed objects, which reduces noisy entries in the final database.
The implemented system includes a confirmed-only semantic layer, an object-to-goal interface for Nav2, onboard Spot deployment, RViz/rosbag tooling, and repeated-run supervision. Evaluation covered three real indoor environments: a church, the DLS lab, and a large IIT test room.
The same pipeline produced usable 2D maps across different layouts and kept the semantic database compact by focusing on a small set of object classes. The main observed limitations were depth alignment bias in object placement and localization inconsistency causing early goal acceptance in edge cases. Both failure modes were visible in logs and RViz, making the system inspectable and technically actionable.
Optimal PID Controller Tuning for Time-Delayed Systems Using Ant Colony Optimization
This project was my Bachelor thesis during my studies at Qazvin Azad University under the supervision of Prof. Hamid Qadiri.
This project investigates PID tuning with Ant Colony Optimization for a second-order control system, while explicitly modeling a delayed baseline to study how time delay affects the transient response. The goal is to minimize the Integral of Time-weighted Absolute Error (ITAE) and obtain a faster, cleaner response with reduced oscillation and smaller tracking error.
We implemented the project in MATLAB and combined classical transfer-function modeling, heuristic optimization, and result visualization. We first build the delayed reference model, then search for PID gains that improve closed-loop performance through repeated simulation and pheromone-based updates.
We define the plant as G = tf([1], [1 2 3]), which corresponds to a
second-order transfer function. A pure delay is then introduced through exp(-1 * s). Since a
pure delay is not directly convenient for rational transfer-function analysis, we use a
first-order Pade approximation to construct the delayed model used for the baseline step response.
Plant transfer function
G(s) = 1 / (s2 + 2s + 3)
Time delay term
e-s
First-order Pade approximation
e-s ≈ (1 - s/2) / (1 + s/2)
Delayed model used for baseline analysis
Gt(s) = G(s) . Pade[e-s]
[1 2 3].exp(-s) with pade(delay, 1) to obtain a rational delayed model.Gf = feedback(Gt, 1) and plotted before controller tuning.The controller follows the standard PID structure with proportional, integral, and derivative gains. The performance target is ITAE, which weights the tracking error by time and therefore penalizes not only large deviations but also errors that persist for too long.
PID controller law
u(t) = Kpe(t) + Ki∫e(t)dt + Kdde(t)/dt
ITAE objective
ITAE = ∫0T t|e(t)|dt
Closed-loop model evaluated in the optimization loop
Gcf(s) = feedback(C(s)G(s), 1)
One implementation detail is worth stating clearly: in our current implementation, the delayed model is
used for the initial baseline plot, while the ACO loop evaluates PID candidates on the nominal plant
through feedback(Gc * G, 1). This page reflects that implementation exactly.
We use 10 ants, 10 iterations, a pheromone decay factor of 0.6, a scaling factor of 2, and three tuning variables for Kp, Ki, and Kd. Candidate gains are sampled through pheromone-weighted roulette-wheel selection.
For each ant, MATLAB builds a controller with pid(kp, ki, kd), simulates the step response, computes an ITAE-style cost, and reinforces the pheromone trail of the best candidate in that iteration while the others decay.
Our MATLAB workflow is organized around four stages: system definition, ACO initialization, iterative search, and result visualization. We define the nominal and delayed models, allocate the pheromone tables and candidate arrays, evaluate sampled PID gains in closed loop, and store the best cost value across iterations.
G, builds the delayed model with pade(delay, 1), and plots the unity-feedback delayed response.
Figure 1. Step response of the original delayed system. The uncompensated response shows overshoot and oscillatory behavior, highlighting the performance degradation introduced by delay.
Figure 2. Step response after tuning the PID gains with ACO. The optimized controller improves settling behavior and produces a noticeably cleaner transient response.
Figure 3. ITAE across optimization iterations. The downward trend indicates convergence toward more effective PID gains.
The project is implemented in MATLAB with the Control System Toolbox. Running the main file builds the delayed baseline model, executes the ACO-based PID search, reports the best gains and minimum ITAE in the command window, and generates the three figures used in the thesis.
PID_ACO.m.kp, ki, kd, and Min_ITAE.This thesis shows that Ant Colony Optimization can be used effectively to tune a PID controller for a second-order control problem with a delayed reference model. By minimizing ITAE, the optimized controller achieves a cleaner transient response and better settling behavior than the uncontrolled baseline.
It also demonstrates the value of combining classical control tools with heuristic optimization in a compact, reproducible MATLAB workflow. The project was a solid first experience in algorithmic controller tuning and technical evaluation during my undergraduate studies.
My experience sits at the intersection of mobile robotics, perception, semantic mapping, autonomous navigation, and practical robotics system development.
Semantic object-goal navigation on quadruped robots, combining perception, mapping, and Nav2-based execution on physical platforms.
Team-based robotics research and development in RoboCup Small Size League, including competitive system integration and documentation.
Applied robotics and engineering collaboration through the University of Genoa ecosystem.
Engineering experience through Radman System and ORBI, with practical implementation work beyond academic projects.
My master's thesis presents a practical object-goal navigation pipeline for Boston Dynamics Spot in real indoor environments. A pre-run records a 2D map and semantic object database; during operation, the robot localizes on the saved map, selects a recorded object target, and navigates using a layered Nav2 costmap with onboard sensing and computation.
This paper describes an online semantic object mapping system for quadruped robots. The mapper combines range geometry with RGB D detections, merges repeated observations into persistent object instances, and produces a compact object layer that can be queried by class, pose, and confidence.
This team description paper presents the MRL Small Size League robot hardware and software design, including rule-driven software updates, reliability improvements, accuracy enhancements, and changes to the electronic structure that enabled more complex robot behavior.
I am open to PhD, research, and robotics engineering opportunities involving autonomy, perception, navigation, SLAM, and real-world robot deployment.