Time Flies: Wayfinder’s One Year Anniversary
It’s hard to believe one year has passed since Wayfinder started. As experts in the innovation of flight you can take it from us - it turns out time really does fly!
Our team has been doing some reminiscing on the past year and as we looked through photos and talked about our favorite memories, we felt this celebration was the perfect opportunity to share some of those special moments with all of you.
“The Wayfinder team is focused on enabling self-piloted aircraft operations. Doesn't that sound so cool? Well, it did to me! This could be the next big thing for the future.” - Palak Chandalia
There are three starting points to be successful in building Machine Learning based autonomy solutions:
Build a strong team with relevant experience, such as algorithm development
Build a high performance infrastructure to aggregate and process large amounts of data
Collect and generate large amounts of relevant data, such as to train and validate Machine Learning models
This past year has been focused on establishing these starting points:
1. Team: Our team more than doubled in a year
2. Infrastructure: We designed and built in-house a state of the art data infrastructure specific to autonomy machine learning needs.
“As an infrastructure person, I'm actually looking forward to running out of resources, having to buy more storage/CPU/GPU, in our on-prem or public cloud infra. From a technology perspective, as we grow our team and the number of services we support, we're headed towards a more flexible infrastructure design with kubernetes, and towards higher-demand, multi-GPU, and multi-node training.” - Romain Komorn
3. Data Collection: We designed, built, and tested on-vehicle data collection devices, made significant progress to generate synthetic data, and continued to collect data on the Vahana aircraft (see our recently released overview video).
“Getting ready for the OWL-CAM first flight was a push - we reached Hayward airport by 9 am and got started on hooking up the equipment to the helicopter, once it was ready to go it was just a matter of waiting. After ten minutes of having the rotors running we finally saw them take off which was a relief, and after the test was complete, we examined the logs that the test went without a hitch with the system!” - Sid Kumar
With an initial team, infrastructure, and data in place, we were able to make great strides in developing solutions, including good progress in our work on ATTOL - Airbus’ Autonomous Taxi, Takeoff, and Landing demonstrator. Our initial experiments gave us a better understanding of the practical complexities and nuances of the problem, such that we are now generating more and better-distributed synthetic datasets with a wider variety of airport and runway layouts, as well as training networks using more consistent cues and metrics for localization.
We also want to thank the opportunity to those who have followed along with our story this past year and hope you continue to read the blog, stay in touch, and even apply to join our team. Romain said it best when he described why he wanted to join the Wayfinder team: “Self-flying planes and taxis are going to change the world and I want to be a part of it.”