IvLabs

Autonomous Rocket Lander - Summer Internship Project 2025
What does it take to teach a rocket how to land itself?
Using Deep Reinforcement Learning (PPO), we trained an AI agent to perform controlled rocket landings by learning from trial and error in a simulated environment.
Key Highlights
Continuous Control
The agent learns to manage throttle, side thrust, and nozzle angle in real time.
Rich State Information
Position, velocity, orientation, angular velocity, and landing-leg contacts are used to make decisions.
Reward Design
Rewards encourage smooth, precise, and fuel-efficient landings while penalizing crashes and unstable behavior.
Multiple Landing Scenarios
The rocket was trained on both stationary and moving barges to improve robustness and adaptability.
Results
From random actions to reliable soft landings, the agent gradually learned stable descent strategies and achieved consistent performance after training.
Deep Reinforcement Learning in action - transforming exploration into intelligent control.
Team Members:
Raj Patil, Tvisha Mehta, Ishan Agrawal
Mentors:
Akash Tiwari, Atharv Kulkarni, Aarush Sinha, Anuj Sidam

Image Processing on FPGA | IvLabs, VNIT Nagpur:

Designed and implemented a real-time FPGA-based image processing system, handling image transfer from a PC, on-chip storage and hardware processing, with output displayed on a VGA monitor.
Instead of software-based processing, all image operations are implemented using Verilog RTL, enabling parallelism, low latency, and real-time performance.

Technical Highlights:
• Image input via UART (RGB444, QVGA 320×240)
• On-chip storage using Dual-Port BRAM
• VGA controller for 640×480 @ 60 Hz output
• Real-time filters using convolution hardware
• Line buffers for 3×3-pixel window operations
• Mode control and filter selection via FPGA switches

Supported Filters:
Grayscale | Inversion | RGB channel selection | Laplacian edge detection | Sobel edge detection

Key Learnings from project:
Hands-on experience with FPGA-based image pipelines, VGA timing, FSM design, and hardware acceleration, bridging digital design and image processing.

Potential Applications:
• Embedded vision systems
• Real-time image processing
• Industrial inspection and monitoring
• Surveillance and security systems
• Robotics and autonomous systems
• FPGA-based learning and research platforms

Team Members –
sgh_
*ta.18_


Team Mentors –
_.aayush

Summer Intern Project 2025 | Gesture Controlled Quadcopter.

For our Summer Intern Group Project, we developed a Gesture controlled Quadcopter which replaced the traditional RC transmitter with a custom gesture-based control model
,by tracking 21 real-time hand landmarks using Mediapipe.

The Tech Stack:

Hardware: Pixhawk Cube Orange + S500 Frame
Intelligence: Mediapipe Landmark Tracking
Communication: MAVLink & Dronekit
Testing: SITL (Software In The Loop) for safety

Learning outcomes:
By mapping real-time hand movements to flight commands, we understood how to bridge the gap between human intent and complex robotics. We successfully validated our logic through SITL simulations and synchronized software-hardware integration for stable flight.
Check out the results!



Team mentors - .123456789
Team members- .59

Ever wondered how to be a part of IVLab? 👀
Technoseason is your gateway 🚀
Rules are simple. Skills matter.
⏳ Registrations open till 5th Feb 2026 ,midnight
Don’t miss out!

🚗🔥 Summer Intern Project 2025: Autonomous Car using Deep Reinforcement Learning 🤖🛣️
In this project, we built an Autonomous Driving System using Deep Reinforcement Learning, where an agent learns to drive a car by interacting with the environment and improving through trial and error , just like a human driver learning from experience.
🧠 What makes it exciting?
Instead of hard-coded rules, the car learns optimal driving behavior by maximizing rewards such as staying on the road, avoiding collisions, and reaching the destination efficiently.
⚙️ Technical Highlights:
• Environment: OpenAI Gym / simulation-based driving setup
• State space: Visual observations
• Action space: Steering, Acceleration, Braking
• Algorithm: PPO ( Proximal Policy Optimization )
• Neural Network: Deep NN ( Actor - Critic )
• Reward design for safe and smooth driving
📊 Outcome:
The agent successfully learned lane-following, obstacle avoidance, and stable driving behavior through continuous interaction with the environment.
🌍 Applications:
Self-driving cars, Robotics Navigation, Traffic Simulation, Smart Mobility, and Autonomous Systems Research.
🚀 Key Learning:
A strong step forward from supervised learning → reinforcement learning → real-world autonomous decision-making, combining perception, control, and intelligence.

👥 Team Members: Taha Motorwala, Ananya Goswami , Pratik Nawale
🎓 Mentors: Aarush Sinha, Atharv Kulkarni , Akash Tiwari , Anuj Sidam
🔗 GitHub: https://github.com/tahamm786/Autonomous-Car-Driving-using-Deep-RL