VERTLINER: Autonomous Robotic Assessments in Indoor Environments
VERTLINER is one of seven winners of the second ELSA Industry Call and has received 60,000 euros in funding. For the duration of the Industry Call, VERTLINER has worked together with top AI experts within our Network of Excellence.
VERTLINER develops autonomous capabilities for aerial platforms, enabling them to conduct autonomous inspection, monitoring, surveillance and/or reconnaissance activities in GPS-denied and non-permissive environments.
Result Video
In this video, the team presents the joint project with ELSA:
Short Summary
“During our collaboration with ELSA, VERTLINER advanced SENSATE, a UAV control system focused on safe, privacy-preserving navigation in GPS-denied and unknown environments. We built a high-fidelity simulation environment and a digital twin of our UAV to prototype and test autonomous navigation behaviors with LiDAR sensing.
Within this setting, we developed and trained a neural network-based controller, trained with reinforcement-learning to drive the agent from start to goal while avoiding obstacles. The policy learned stable, collision-averse behavior which was validated in multiple simulation scenarios, indicating improved safety and more predictable actions.”
“To narrow the sim-to-real gap, we modelled the real UAV’s physical parameters within the simulator, preparing for controlled data collection and field trials.”
“Next steps defined under the project include:
- fine-tuning with real-world demonstration data,
- introducing human-in-the-loop feedback and reward shaping, and
- validating performance on real tasks before compiling the final report.
Collectively, these activities strengthened our capability for autonomous exploration, obstacle avoidance, and mapping in constrained environment while maintaining a general, publishable scope.”
Learn more about VERTLINER: