syllabus.md

layout	page
title	Syllabus
description	Listing of course modules and topics.

Course Information:

Field	Details
Course	CSE 468/568: Robotics Algorithms
Term	Fall 2026
Meetings	Tuesday/Thursday, 2:00–3:20 pm
Location	Knox 110
Instructor	Karthik Dantu
Email	kdantu@buffalo.edu
Phone	(716) 645-2670
Website	https://www.cse.buffalo.edu/faculty/kdantu
Office	Davis 113F
Office hours	Monday, 2:00–3:00 pm; Wednesday, 4:00–5:00 pm

Course Description: This course is intended to be a comprehensive introduction to robotics algorithms for a senior undergraduate/first-year graduate student. It is a Computer Science course, and introduces the student to well-known algorithms in making a simple robot autonomous.

Course Format: Sixteen weeks; two 90-minute meetings per week. A standard meeting uses 15 minutes for retrieval and focused derivation, 60 minutes for guided and independent implementation, 10 minutes for a demonstration or failure analysis, and 5 minutes for a commit and exit note. The running platform is Linux, ROS 2, deterministic simulation, recorded bags, and an F1TENTH-class vehicle when available.

Prerequisite(s): Data structures and algorithms; introductory probability; linear algebra and calculus; programming in Python, C++, or Java. Prior robotics is helpful, but not required.

Credit Hours: 3 (468)/ 3 (568)

Textbooks and Readings

There are no required textbooks for this class. Material is selected from a collection of reference books. They include:

Introduction to Autonomous Mobile Robots, 2^nd^ Edition
Author(s): Ilah Nourbaksh, Ronald Seigward, Davide Scaramuzza
ISBN-13: 978-0262015356

Probabilistic Robotics
Author(s): Sebastian Thrun, Wolfram Burgard, Dieter Fox
ISBN-13: 978-0262201629

Introduction to Autonomous Robots
Authors: Correll, Nikolaus and Hayes, Bradley, and Heckman, Christoffer, and Roncone, Alessandro
Freely available online in the link provided.

Planning Algorithms
Authors: Steve Lavelle
Freely available online.

The primary text for the topic-by-topic readings below is Introduction to Autonomous Robots: Mechanisms, Sensors, Actuators, and Algorithms.

Topic-by-Topic Course Readings:

The readings identify the relevant chapter, section, subsection, and page range in the primary text.

• Autonomy systems and actuation
  • Autonomy architecture and mobile-robot challenges — Ch. 1, §§1.3–1.4 (pp. 14–16).
  • ROS 2, simulation, bags, TF, logging, and reproducibility — course software material; Ch. 1, §1.3 provides autonomy-system context (pp. 14–15).
  • Reactive control — Ch. 11, §11.1 (pp. 193–196).
  • Finite-state machines, behavior trees, and mission planning — Ch. 11, §§11.2–11.5.1 (pp. 197–207).
  • Electric motors, motor controllers, and actuator safety — Ch. 6, §§6.1–6.3 (pp. 105–112).
• Robot geometry and motion
  • Locomotion, stability, and degrees of freedom — Ch. 2, §§2.1–2.3 (pp. 25–33).
  • Coordinate frames and homogeneous transformations — Ch. 2, §§2.4.1–2.4.3 (pp. 35–40).
  • Euler angles and quaternions — Ch. 2, §2.4.4 (pp. 40–42).
  • Differential-drive kinematics — Ch. 3, §3.3.2 (pp. 59–65).
  • Odometry — Ch. 3, §3.3.2, “From Forward Kinematics to Odometry” (pp. 65–66).
  • Car-like/Ackermann kinematics — Ch. 3, §3.3.3 (pp. 66–67).
• Control, tracking, and learning
  • Feedback control and PID — Ch. 3, §3.4.2 (p. 72); PID is developed in course material.
  • Trajectory tracking — Ch. 3, §3.4.2 (p. 72); Pure Pursuit and Stanley tracking are developed in course material.
  • LQR, MPC, trajectory optimization, and basic reinforcement learning — course material; neural-network foundations in Ch. 10, §§10.1–10.5 (pp. 170–182).
• Sensors and robot perception
  • Sensor terminology, error sources, and sensor classes — Ch. 7, §§7.1–7.1.1 (pp. 119–120).
  • Encoders, accelerometers, gyroscopes, and range sensing — Ch. 7, §§7.2–7.5.3 (pp. 121–130).
  • Image processing and camera-derived structure — Ch. 8, §§8.1–8.4 (pp. 139–151).
  • LiDAR algorithms: line fitting and RANSAC — Ch. 9, §§9.3.1–9.3.3 (pp. 157–161).
  • Feature descriptors and object recognition — Ch. 9, §§9.4–9.5 (pp. 162–165).
  • Neural networks, CNNs, and recurrent networks — Ch. 10, §§10.1–10.7 (pp. 170–188).
  • Deep-learning perception, object detection, segmentation, and vision-language perception — course material; Ch. 10, §§10.3–10.7 provides foundations (pp. 175–188).
  • ICP, voxel mapping, and RGB-D mapping — Ch. 12, §§12.2–12.4 (pp. 214–219).
• Probability, estimation, and localization
  • Probability, uncertainty propagation, sensor fusion, and Kalman filtering — Ch. 15, §§15.1–15.3.1 (pp. 258–266) and Appendix C, §§C.1–C.5 (pp. 321–328).
  • Markov localization and Bayes filters — Ch. 16, §§16.2–16.3.1 (pp. 271–281).
  • Particle filters and EKF localization — Ch. 16, §§16.4–16.5.1 (pp. 281–287).
• Mapping and SLAM
  • Map representations and dense mapping — Ch. 12, §§12.1–12.4 (pp. 213–219).
  • Landmark-based and graph-based SLAM — Ch. 17, §§17.1–17.4.2 (pp. 292–303).
  • Learned SLAM, semantic maps, and visual-language-action systems — course material; graph-SLAM foundations in Ch. 17, §§17.3–17.4.2 (pp. 295–303).
• Planning and navigation
  • Configuration spaces, Dijkstra’s algorithm, and A* — Ch. 13, §§13.1–13.2.2 (pp. 224–228).
  • Costmaps and diffusion-based planning — course material; map-representation foundations in Ch. 12, §12.1 (pp. 213–214).
  • RRT, planning at multiple scales, and coverage planning — Ch. 13, §§13.3–13.5 (pp. 228–235).
  • PRM/RRT*, kinodynamic planning, Dynamic Window, and trajectory rollout — course material; RRT foundations in Ch. 13, §13.3.1 (pp. 229–233).
• Deployment and responsibility
  • Safety, privacy, cybersecurity, human interaction, sustainability, and responsible deployment — Ch. 6, §6.3 (p. 112) and Appendix C, §§C.5–C.5.4 (pp. 325–328), with course material on fault injection and safety monitors.

Course Objectives:

At the completion of this course, students will be able to:

1. Kinematics. Model and control a mobile robot by applying coordinate-frame reasoning, kinematics, feedback, trajectory tracking, and appropriate assumptions about sensing, timing, and actuation.
2. Perception. Design and evaluate robot-perception pipelines that combine range, inertial, camera, and learned visual information while accounting for calibration, uncertainty, and failure modes.
3. Estimation. Implement and assess probabilistic state-estimation, mapping, localization, and SLAM algorithms using quantitative evidence from deterministic simulation and recorded data.
4. Planning. Formulate and compare navigation and decision-making methods, including graph search, sampling-based planning, local planning, optimization, reinforcement learning, and learned autonomy components.
5. ROS. Architect, implement, test, and debug an integrated autonomy stack in ROS 2 with clear subsystem interfaces, reproducible experiments, safety monitors, and measured system behavior.
6. Security and Ethics. Exercise professional judgment in the deployment of autonomous robots by communicating results clearly and considering safety, security, privacy, ethics, human interaction, and societal impact.

While this is a reasonable outline, the instructor reserves the right to change the syllabus as he sees fit. Please check the course site for the latest changes.

Grade Distribution:

Component	Weight	Grading basis
Programming assignments	30%	Five cumulative assignments: PA1, PA2, PA3, and PA4 are 5% each; PA5 is 10%. Correctness, evidence, robustness, and explanation are evaluated.
Implementation activities	20%	Approximately 12–14 low-stakes activities, each worth 2%. The ten highest activity scores count; the lowest two scores are dropped when 12 activities are offered, the lowest three when 13 are offered, and the lowest four when 14 are offered. Activities include traces, tests, and diagnostic artifacts and stop short of an assignment’s main integration.
Quizzes	30%	Short quizzes assess conceptual understanding, derivation, code reading, prediction, debugging, and interpretation of results.
Midterm examinations	20%	Midterm examination(s), collectively worth 20%, assess integrated understanding of robot models, algorithms, assumptions, and failure modes.

Grades will be on a curve. From prior experience, I have seen different performance from the undergraduate (468) and graduate (568) students. Therefore, for fairness, I'll grade students on two separate curves, one for graduate students and another for the undergraduate students.

Course Policies:

• General: No makeup quizzes or in-class activities will be given unless discussed on a per case basis well in advance.

• Grades

  • Graded on a curve with B+ roughly being the median.

  • Grades will be maintained in the myUB course system. Students are responsible for tracking their progress by referring to the online gradebook and reporting any discrepancies.

• Labs and Assignments

  • Students are expected to work independently unless an assignment expressly permits collaboration. Offering and accepting solutions from others outside of permitted collaboration is an act of plagiarism, which is a serious offense and all involved parties will be penalized according to the Academic Honesty Policy. Discussion amongst students is encouraged, but when in doubt, direct your questions to the professor, teaching assistant, or graders.

  • Unless addressed with the professor well ahead of a deadline, late submissions will not be allowed.

• Attendance and Absences

  • Attendance is expected each class. Activities and quizzes are in-class. The professor will not entertain any grade changes toward the end of the course if the student has not participated during the semester.

  • If you have a medical reason for missing class, I will require a doctor's note. Without clear directions from a doctor asking time off for you, I will assume you are able to do the class work and attend class.

  • Students are responsible for all missed work, regardless of the reason for absence. It is also the absentee's responsibility to get all missing notes or materials.

Plagiarism Policy:

• This course has several programming assignments and in-class activities mostly based on the ROS 2 programming environment. We use sophisticated code checkers to check for code copied from assignments from this class as well as submissions from prior editions of the class. It is very improbable that you will be able to fool the code checker.

• Along the same lines, if you see someone else's code to understand the logic, it is probable that our code checker will flag this as plagiarized since your code will be influenced by what you saw and will look structurally similar.

• Discussing high-level programming ideas with other students is acceptable. However, this is a slippery slope, and the more detailed these discussions are and the more they tend toward actual code, the more likely the code checker will flag this as plagiarism.

• Copying snippets of code from online resources (such as StackOverflow, GitHub, etc.) is also considered plagiarism. When you are in doubt, please check with the professor for clarity.

• The AI-assisted programming workflow is described in Operating Expectations below. It does not override the individual-work, permitted-collaboration, or academic-integrity requirements above.

Course Schedule:

Date	Topics	Activities	Assignments
Aug. 27 (Thu.)	Autonomy architecture and mobile-robot challenges	—	—
Sept. 1 (Tue.)	ROS 2, simulation, bags, TF, logging, and reproducibility	A01 assigned	—
Sept. 3 (Thu.)	Reactive control	A01 due; A02 assigned	—
Sept. 7 (Mon.)	Labor Day holiday observed — no classes	—	—
Sept. 8 (Tue.)	Coordinate frames and homogeneous transformations	A02 due; A03 assigned; A03 due Sept. 13	—
Sept. 10 (Thu.)	Differential-drive kinematics; odometry; car-like/Ackermann kinematics	—	—
Sept. 15 (Tue.)	Feedback control and PID	A04 assigned; due Sept. 20	PA1 assigned
Sept. 17 (Thu.)	Trajectory tracking	—	—
Sept. 22 (Tue.)	LQR, MPC, trajectory optimization, and basic reinforcement learning	A05 assigned; due Sept. 27	—
Sept. 24 (Thu.)	Sensor terminology, error sources, and sensor classes; encoders, accelerometers, gyroscopes, and range sensing	—	—
Sept. 29 (Tue.)	LiDAR algorithms: line fitting and RANSAC	A06 assigned; due Oct. 4	PA1 due; PA2 assigned
Oct. 1 (Thu.)	Image processing and camera-derived structure; feature descriptors and object recognition	A07 assigned; due Oct. 11	—
Oct. 6 (Tue.)	Neural networks, CNNs, and recurrent networks	—	—
Oct. 8 (Thu.)	Deep-learning perception, object detection, segmentation, and vision-language perception	—	—
Oct. 12–13	Fall Break — no classes	—	—
Oct. 15 (Thu.)	ICP, voxel mapping, and RGB-D mapping	—	PA2 due; PA3 assigned
Oct. 20 (Tue.)	Probability, uncertainty propagation, sensor fusion, and Kalman filtering	A08 assigned; due Oct. 25	—
Oct. 22 (Thu.)	Markov localization and Bayes filters	A09 assigned; due Nov. 1	—
Oct. 27 (Tue.)	Particle filters and EKF localization	—	—
Oct. 29 (Thu.)	Map representations and dense mapping	A10 assigned; due Nov. 8	PA3 due
Nov. 3 (Tue.)	Landmark-based and graph-based SLAM	—	PA4 assigned
Nov. 5 (Thu.)	Learned SLAM, semantic maps, and visual-language-action systems	—	—
Nov. 10 (Tue.)	CoRL travel day — no class	—	—
Nov. 12 (Thu.)	CoRL travel day — no class	—	—
Nov. 17 (Tue.)	Configuration spaces, Dijkstra’s algorithm, and A*	A11 assigned; due Nov. 22	PA4 due; PA5 assigned
Nov. 19 (Thu.)	Costmaps and diffusion-based planning	A12 assigned; due Dec. 1	—
Nov. 24 (Tue.)	RRT, planning at multiple scales, and coverage planning	A13 assigned; due Dec. 3	—
Nov. 25–28	Thanksgiving Break — no classes	—	—
Dec. 1 (Tue.)	PRM/RRT*, kinodynamic planning, Dynamic Window, and trajectory rollout	A12 due; A14 assigned; due Dec. 3	—
Dec. 3 (Thu.)	Safety, privacy, cybersecurity, human interaction, sustainability, and responsible deployment	A13 due; A14 due	—
Dec. 7 (Mon.)	Official Race and Demo Day	—	—
Dec. 8 (Tue.)	Reading Day	—	PA5 due
Dec. 9–16	Final examinations	—	—

Programming Activities:

These short implementation activities build individual components that support the programming assignments. They adapt the course’s ROS, control, perception, estimation, and planning exercises to the current autonomy stack.

1. A01 — ROS 2 Demo. Run a minimal ROS 2 publisher/subscriber demo and identify the nodes, topics, and message flow.
2. A02 — ROS 2 Basics 1. Create a node that publishes, subscribes, and reports a robot-state value through the ROS 2 graph.
3. A03 — ROS 2 Basics 2. Use launch files, TF, and a recorded bag to reproduce and inspect a short autonomy run.
4. A04 — Control Robot Goal Seek. Implement a goal-seeking controller and document its position and heading error over a run.
5. A05 — Car Pure Pursuit. Tune a Pure Pursuit tracker for a car-like robot and compare tracking performance for two look-ahead choices.
6. A06 — Laser Processing and TurtleBot RANSAC. Filter a LiDAR scan, estimate a geometric line with RANSAC, and report inliers plus one failure case.
7. A07 — Feature Matching and Alignment. Detect, match, and visualize image features; identify a correspondence failure case.
8. A08 — Bayesian Estimation. Implement a prediction/correction update and interpret the effect of motion and sensor uncertainty.
9. A09 — Point-Cloud Registration. Register two point clouds or scans and report the estimated transform and residual error.
10. A10 — Occupancy-Grid Map Representation. Convert range observations and robot poses into an occupancy grid; visualize free, occupied, and unknown cells and explain one resolution or sensor-model tradeoff.
11. A11 — Graph Planning — Forward Search. Build a costmap-based forward search with Dijkstra’s algorithm or A*, then reconstruct the path and identify one infeasibility case.
12. A12 — Nav2 Navigation Stack. Configure and exercise Nav2 with a supplied map and robot model; inspect the global plan, local behavior, and one recovery or parameter-setting decision.
13. A13 — Gap Follow and Local Recovery. Implement a Follow-the-Gap or comparable local-recovery behavior and evaluate it in a constrained scene.
14. A14 — Sampling Planner and Safety Watchdog. Generate feasible sampled trajectories and add a simple confidence, collision, or watchdog gate before execution.

Programming Assignments:

The five programming assignments build a cumulative mobile-robot autonomy stack.

1. PA1 — Drive-by-Wire, Frames, and AEB. ROS 2 bring-up, odometry, TF, logging, and a braking supervisor.
2. PA2 — Reactive Driving: See, Follow, and Avoid. LiDAR processing, PID wall following, Follow the Gap, and a compact visual cue.
3. PA3 — Where Am I? Localization, Mapping, and SLAM. Occupancy maps, Monte Carlo localization, and a measured SLAM study.
4. PA4 — Plan the Line, Track the Line. Costmaps, graph/sampling planning, tracking, replanning, and recovery.
5. PA5 — Integrated Autonomy Challenge. Monitored full-stack autonomy plus one advanced extension: learning, opponent perception, or advanced control.

PA5 pairs use named subsystem ownership, shared integration milestones, and an individual checkoff.

Every assignment submission is code plus one short document: what was built, how to run it, one design choice, evidence it works, and one failure case with its likely cause. CSE 568 uses the same core with an added derivation, research-grounded alternative, ablation, consistency/complexity analysis, or statistical comparison.

Operating Expectations:

Every assignment submission is code plus one short document: what was built, how to run it, one design choice, evidence it works, and one failure case with its likely cause. Students use deterministic simulation and recorded bags as the graded fallback. The simulator-to-hardware safety gate remains mandatory for any hardware work. AI-assisted programming is taught through specification, code reading, prediction, testing, diagnosis, repair, and re-test; generated code is treated as untrusted, and no separate AI-use statement is required.

ABET Student Outcomes:

This course supports the following ABET Computing Accreditation Commission (CAC) Computer Science student outcomes.

1. ABET CAC SO 1 — Analyze complex computing problems and apply principles of computing and other relevant disciplines to identify solutions. Direct contribution: Students formulate sensing, estimation, planning, control, and integration problems; select representations, algorithms, and assumptions; and diagnose failures in deterministic replay and simulation.
2. ABET CAC SO 2 — Design, implement, and evaluate a computing-based solution to meet a given set of computing requirements in the context of the program’s discipline. Direct contribution: PA1–PA5 require students to specify interfaces and constraints, implement robot-autonomy components, test them against scenario requirements, and evaluate the integrated solution with evidence.
3. ABET CAC SO 3 — Communicate effectively in a variety of professional contexts. Reinforced contribution: Each assignment requires a concise, reproducible technical document that states what was built, how to run it, one design choice, evidence it works, and one failure case with its likely cause; demonstrations and peer reproduction reinforce technical communication.
4. ABET CAC SO 4 — Recognize professional responsibilities and make informed judgments in computing practice based on legal and ethical principles. Direct contribution: Students evaluate safety, privacy, security, accessibility, environmental, and societal implications of autonomy-system design choices. The intended direct-assessment artifact is an individual written design-decision exercise scored on three performance indicators: recognition of professional responsibilities, informed judgment based on legal principles, and informed judgment based on ethical principles.
5. ABET CAC SO 5 — Function effectively as a member or leader of a team engaged in activities appropriate to the program’s discipline. Reinforced contribution: PA5 pairs use named subsystem ownership, shared integration milestones, and an individual checkoff; students plan tasks, coordinate interfaces, and meet common evaluation objectives.
6. ABET Computer Science SO 6 — Apply computer science theory and software development fundamentals to produce computing-based solutions. Direct contribution: Students implement and test algorithms for transforms, kinematics, filtering, mapping, planning, control, and learning; the course emphasizes modular software, algorithmic complexity, reproducibility, debugging, and regression tests.

Assessment Evidence: Candidate evidence includes assignment repositories and tests, fixed-seed simulation results, bag-replay traces, quantitative comparisons, concise technical documents, individual safety/ethics decision exercises, and PA5 integration demonstrations. CSE 568 students provide the same core evidence with an added derivation, research-grounded alternative, ablation, consistency/complexity analysis, or statistical comparison.