Skip to content

Commit

Permalink
update doc
Browse files Browse the repository at this point in the history
  • Loading branch information
mrunaljsarvaiya committed Oct 12, 2024
1 parent 1ad71f1 commit d4d887a
Show file tree
Hide file tree
Showing 4 changed files with 19 additions and 24 deletions.
24 changes: 11 additions & 13 deletions _projects/1_peanut_project.md
Original file line number Diff line number Diff line change
Expand Up @@ -20,17 +20,15 @@ Peanut Robotics builds mobile manipulators for autonomous cleaning and disinfect
</div>


During my 4 years at Peanut, I built and integrated a number of core features for our autonomy stack including but not limited to

- Developed the motion planning software architecture for a custom mobile manipulator
- Selected, integrated and tuned a trajectory tracking controller and time optimal trajectory generator (TOPPRA)
- Designed and implemented a custom navigation planner for hotel hallways that generates human-like and aesthetically pleasing motions
- Trained a neural network model to estimate manipulator currents used to add trajectory generator torque constraints for enhanced reliability
- Engineered tools to optimize robot paths by smoothing trajectories and reducing
overall joint-space distance by tweaking input configurations using an optimization program
- Communication architecture between controllers and the main on-board computer
- Error recovery and safety features for the mobile base motors
- State machines for executing missions during deployment
- RViz-based tool to record, modify and playback cartesian paths
- Collision checking tools (through MoveIt) used by other planners in the stack
During my 4 years at Peanut, I built and integrated a number of core features for our autonomy stack:

- Developed the motion planning software architecture for a custom-built mobile manipulator.
- Selected, integrated, and fine-tuned a trajectory tracking controller along with a time-optimal trajectory generator (TOPPRA).
- Designed and implemented a custom navigation planner for hotel hallways, generating human-like and aesthetically pleasing motions.
- Trained a neural network model to estimate manipulator currents, adding torque constraints to the trajectory generator for improved system reliability.
- Engineered optimization tools to smooth robot paths, minimize joint-space distances, and adjust input configurations using advanced optimization techniques.
- Developed the communication architecture between controllers and the main onboard computer, ensuring efficient data exchange.
- Implemented error recovery mechanisms and safety features for mobile base motors to enhance operational robustness.
- Created state machines for mission execution during deployment, streamlining task automation.
- Designed an RViz-based tool for recording, modifying, and replaying Cartesian paths, aiding in testing and debugging.
- Integrated collision-checking tools via MoveIt to support other planners in the robotic stack.
15 changes: 6 additions & 9 deletions _projects/2_sae_project.md
Original file line number Diff line number Diff line change
Expand Up @@ -15,8 +15,7 @@ related_publications: false

After leading projects on the engine subsystem, I developed the skills required to efficiently manage a team and work cooperatively with a group of engineers.

I moved onto leading the administrative and external aspects of the team as Team President. As President I was responsible for organizing and holding team meetings for over 60 students, both general as well as team leader meetings. My duties extended to managing the team budget and allocating the funds received across 6 subsystems while keeping a record of all trascations.

I then transitioned to leading the administrative and external functions of the team as Team President. In this role, I organized and facilitated meetings for over 60 members, including both general sessions and leader-specific meetings. Additionally, I managed the team’s budget, overseeing the allocation of funds across six subsystems and maintaining detailed records of all transactions.

As the external image of the team I developed and maintained relationships with corporate sponsors and the university. To increase public awareness about our team and the field of engineering in general, I organized and held external events such as outreach events at highschools, engineering openhouse booths and student recruiting events at the university.

Expand All @@ -29,9 +28,9 @@ I succeeded in increasing student involvement from majors outside the engineerin
{% include figure.liquid loading="eager" path="assets/img/fsae/fsae_sales.jpg" title="example image" class="img-fluid rounded z-depth-1 uniform-image-height" %}
</div>
</div>
For this event each team makes a formal presentation to convince a panel of judges to invest in their race car business proposal. In this role play, the judges typically act as representatives of a fictitious manufacturing company. The competitors in this event will be judged on their ability to create and deliver a business case that convinces the judges that the team’s design best meets the demands of the amateur weekend competition market and that it can be profitably manufactured and marketed. The judges then score the presentation against predefined criteria , including the presentation content, quality of the presentation, its delivery, and the ability to defend the business proposal in the question and answer session.
For this event, each team delivers a formal presentation aimed at convincing a panel of judges to invest in their race car business proposal. In this role-play scenario, the judges act as representatives of a fictitious manufacturing company. Teams are evaluated on their ability to create and present a compelling business case that demonstrates how their design meets the demands of the amateur weekend competition market, while also being profitable to manufacture and market. The judges assess the presentations based on predefined criteria, including content, presentation quality, delivery, and the team’s ability to defend their proposal during the Q&A session.

I was one of the two presenter for this event and was also actively involved in making the presentation and the market research behind it. Primarily I was responsible for coming up with a manufacturing scheme showing how and where all the products would be manufactured along with the associated machinery. I also gathered data to determine the yearly target sales and potential ways to convince buyers to pick our race car over others.
I was one of two presenters for this event and was actively involved in creating the presentation and conducting the market research. My primary responsibility was to develop a manufacturing plan detailing how and where all components would be produced, as well as the necessary machinery. Additionally, I gathered data to estimate annual target sales and identified strategies to persuade potential buyers to choose our race car over competing options.

<div class="row d-flex justify-content-center">
<div class="col-sm-13 mt-6 mt-md-0 d-flex justify-content-center">
Expand All @@ -48,10 +47,8 @@ My co-presenter and I have presented at 4 competitions so far. We have done extr
{% include figure.liquid loading="eager" path="assets/img/fsae/fsae_sim.jpg" title="example image" class="img-fluid rounded z-depth-1 uniform-image-height" %}
</div>
</div>
The intake and exhaust geometries of the car were designed about 5 years ago for a Honda F4i engine. After switching to a Yamaha R6 engine, there was a need to revalidate/redesign the intake and exhaust geometries of the car to optimize them for this new engine. GT-Power is a 2D simulations software that can be used to simulate an engine. The intake and exhaust geometries are modelled in to account for their effect on airflow and consequently their effect on the engines power output, efficiency, etc.By accurately modelling the engine and its associated components, each parameter can be altered to gauge how it affects the engine’s power output. Hence, the goal of this project was to observe how changing the intake’s runner length and exhaust tube lengths would affect the engine. Before parameter sweeps can be performed, the power curve predicted by GT-Power is to be compared to experimental data through the dynamometer to check the software’s accuracy.
The intake and exhaust geometries of the car were originally designed around five years ago for a Honda F4i engine. After transitioning to a Yamaha R6 engine, it became necessary to revalidate and redesign these geometries to optimize them for the new engine. GT-Power, a 2D simulation software, was used to model the engine, including the intake and exhaust geometries, to evaluate their effects on airflow and, ultimately, the engine's power output and efficiency. By accurately modeling the engine and its components, parameters such as intake runner length and exhaust tube lengths can be adjusted to observe their impact on engine performance. The goal of this project was to determine how altering these geometries would affect the engine’s power output. Before conducting parameter sweeps, we compared the power curve predicted by GT-Power with experimental data obtained from a dynamometer to validate the software’s accuracy.

Simulation results indicated that changes to the exhaust length had little effect on the engine’s power curve. However, adjustments to the intake runner length showed a noticeable impact on power output. As a result, the runner length was increased from 160 mm to 220 mm, optimizing power for the car's average RPM range during autocross and endurance events.

Simulations showed that changing exhaust lengths would not have any significant effect on the engine’s power curve. As for the intake, changing the runner length had a visible effect on the engine’s power output and it was changed from 160 mm to 220 mm to produce more power at the average RPM range of the car during autocross and endurance events.

I am currently working on coupling Star CCM+ with the GT Power model to account for the complex 3D shapes of the intake manifold with a few other members. An Accurate model using Star will more strongly validate the accuracy of the model and allow analyzes different intake manifold designs as well.
Currently, I am working with a team to couple Star CCM+ with the GT-Power model to account for the complex 3D shapes of the intake manifold. This coupling will enhance the accuracy of the model and enable analysis of various intake manifold designs.
Original file line number Diff line number Diff line change
Expand Up @@ -3,7 +3,7 @@ layout: page
title: Squishy Robotics
description: Robotics Software Engineering Intern
img: assets/img/squishy/squishylogo2.jpg
importance: 4
importance: 3
category: work & fun
related_publications: false
---
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -3,7 +3,7 @@ layout: page
title: Stanford AI Graduate Program
description:
img: assets/img/stanford/logo3.png
importance: 3
importance: 4
category: work & fun
related_publications: false
---
Expand Down

0 comments on commit d4d887a

Please sign in to comment.