Rene Mai

Rene Mai

she/her

Mechanical engineering PhD student

Sitemap

A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.

Pages

Posts

Mid-year updates for 2025

less than 1 minute read

Published:

Updates abound! I have been sadly remiss in updating this website, but I am making an attempt to bring it up to speed now. The last year was very eventful, and I’m honored to provide every update below.

Submission to 5th IFAC Workshop on Cyber-Physical Human Systems accepted

less than 1 minute read

Published:

Publication update: I am happy to announce that my submission to the 5th IFAC Workshop on Cyber-Physical Human Systems (2024), “Analysis of human steering behavior differences in human-in-control and autonomy-in-control driving,” has been accepted!

Newly-submitted publication update

less than 1 minute read

Published:

Publication update: Friday June 7, 2024 I submitted a new publication, entitled “Analysis of human steering behavior differences in human-in-control and autonomy-in-control driving,” submitted to the 5th IFAC Workshop on Cyber-Physical Human Systems (2024), with co-authors Agung Julius and Sandipan Mishra. Read more about it on my publications page.

portfolio

publications

Human-as-advisor in the loop for autonomous lane-keeping

American Controls Conference, 2023

This paper presents a human-as-advisor architecture for shared human-machine autonomy in dynamic systems. In the human-as-advisor architecture, the human provides suggested control actions to the autonomous system; the system uses a model of the human controller to ascertain the system’s state as perceived by the human. The system combines this information with additional sensor measurements, yielding an improved state estimate. We apply this architecture to the problem of lane-centering an autonomous vehicle in the presence of conflicting lane markings that render the true lane center uncertain.

Recommended citation: R. Mai, S. Mishra, and A. Julius. (2023). "Human-as-advisor in the loop for autonomous lane-keeping." 2023 American Control Conference (ACC). IEEE, May 31, 2023. doi: 10.23919/acc55779.2023.10156374.
Download Paper

Generalized two-point visual control model of human steering for accurate state estimation

ASME Letters in Dynamic Systems and Control (joint submission to the 2024 Modeling, Estimation, and Controls Conference) (Finalist for Best Student Research Award), 2024

We derive and validate a generalization of the two-point visual control model, an accepted cognitive science model for human steering behavior. The generalized model is needed as current steering models are either insufficiently accurate or too complex for online state estimation.

Recommended citation: Mai, R. E., Sears, K., Roessling, G., Julius, A., and Mishra, S. (October 11, 2024). "Generalized Two-Point Visual Control Model of Human Steering for Accurate State Estimation." ASME. Letters Dyn. Sys. Control. January 2025; 5(1): 011004. https://doi.org/10.1115/1.4066630
Download Paper

Analysis of human steering behavior differences in human-in-control and autonomy-in-control driving

IFAC Workshop on Cyber-Physical Human Systems, 2024

Steering models (such as the generalized two-point model) predict human steering behavior well when the human is in direct control of a vehicle. In vehicles under autonomous control, human control inputs are not used; rather, an autonomous controller applies steering and acceleration commands to the vehicle.

Recommended citation: R. Mai, A. Julius, and S. Mishra. (2024). "Analysis of human steering behavior differences in human-in-control and autonomy-in-control driving." 2024 Workshop on Cyber-Physical Human Systems (CPHS).
Download Paper

Modeling human-autonomy team steering behavior in shared-autonomy driving scenarios

12th IFAC Symposium on Intelligent Autonomous Vehicles, 2025

Well-accepted models such as the two-point steering model and its variations describe human steering behavior in non-autonomous vehicles. However, these models may not describe human steering in a shared autonomous vehicle, where the human driver cooperates with an autonomous controller. This work explores how the generalized two-point steering model, a variation of the classical two-point model, may apply to human steering in a shared autonomous vehicle.

Recommended citation: R. Mai, K. Daveron, A. Julius, and S. Mishra. (2025). "Modeling human-autonomy team steering behavior in shared-autonomy driving scenarios." 2025 IFAC Symposium on Intelligent Autonomous Vehicles.
Download Paper

Performance measures and sim-to-real gap assessment of human-autonomy teaming in obstacle avoidance

ACM SigSim PADS PhD Colloqium, 2025

This paper proposes a study of human-autonomy teaming for obstacle avoidance in a realistic, rich environment. We propose several metrics to measure the overall team performance. We also propose validating how well programmed human behavior based on validated cognitive science models can imitate human participants.

Recommended citation: R. Mai, A. Julius, and S. Mishra. "Performance measures and sim-to-real gap assessment of human-autonomy teaming in obstacle avoidance" 2025 SigSim PADS PhD Colloquium.
Download Paper

talks

Human-as-advisor for human-in-the-loop control

Published:

Presented the human-as-advisor approach to shared autonomy, which uses human-suggested control actions to infer system state information via a human stering model.

Intellectual Property Rights, IP Litigation, Economics, and Innovation

Published:

Overview of US intellectual property law: copyright, trademark, trade secret, and patent law. Discussion of the economic and competitive benefits of intellectual property protection for companies and individuals, as well as the impact of intellectual property claims on individual economic decisions like cell phone purchases. Concluded with a question and answer session about legal careers and intellectual property protections for burgeoning entrepreneurs.

Building a pipeline for large language model fine-tuning, with a semantic search application

Published:

Presented project building a pipeline for document ingestion and encoding as part of the process for preparing to fine-tune a large language model for aviation-specific applications. Project demonstrated semantic search over documents, ensuring that responses are grounded in truth and preventing hallucination. Demonstrated dramatic performance improvements over state of the art language models under aviation-specific questioning.

Introduction to reinforcement learning

Published:

Lecture covers basic reinforcement learning concepts, starting from the basics of state, action, policy, and reward, then introducing the Bellman equation. Students are introduced to the differences between value iteration and Q learning and the relative advantages of each, along with simple discussion of how the problems for each method are tackled. Students simulate a reinforcement learning task in a physical, interactive exercise.

Academic ethics and responsible studying in the age of ChatGPT

Published:

Students are guided through pre-discussion learning via worksheets. Discussion is socratic style, to introduce students to common forms of discussion classes and prompt them to deeply consider the questions LLMs require developers and users to answer for responsible development, deployment, and use. Examples of LLM hallucination are used to illustrate the importance of correct prompt choices and critical reading. Students learn the nuances of academic ethics when using LLMs. Lesson concludes with an exercise in autoregression that puts the students in the place of an LLM responding to a prompt.

Intellectual Property Rights, IP Litigation, Economics, and Innovation

Published:

Overview of US intellectual property law: copyright, trademark, trade secret, and patent law. Discussion of the economic and competitive benefits of intellectual property protection for companies and individuals, as well as the impact of intellectual property claims on individual economic decisions like cell phone purchases. Interactive exercise on patent claim drafting introduces students to the particularities of legal language and the difficulty of describing specific items and phenomena with clear, simple words.

teaching

Graduate teaching assistant, Fall 2022

Undergraduate course, Rensselaer Polytechnic Institute, MANE Department, 2022

Modeling & Control of Dynamic Systems is a required junior-level undergraduate course for all MANE students at RPI. As a required junior-level course, it is one of the largest courses, and proficiency with the subject matter is required for graduation.

Graduate teaching assistant, Spring 2024

Undergraduate course, Rensselaer Polytechnic Institute, MANE Department, 2024

TA for Mechatronics under Professor Sandipan Mishra. Mechatronics is a senior-level elective course that combines the sheer mathematical fun of control theory, the joy of characterizing and using hardware, and the headaches of hardware-software integration into a high-workload course that challenges even the few graduate students who take it. It is, without a doubt, my favorite course to TA.

New Visions STEM Guest Lecturer

High school course, Questar III Boces, 2024

New Visions STEM is a program for bright, ambitious high school students who aspire to careers in STEM with a substantially complete high school education and a desire to challenge themselves.