Nikolaas Bender

Capable Octopus

This is project Capable Octopus, specifically a package called Sidewalk Sculpin. This whole project was for an advanced robotics seminar class where my team was given free reign over what robotics project we wanted to do. We chose to use a Clearpath Jackal due to its availability and as a sensor suite we exclusively used Intel Realsense camera. For navigation we used a neural network running a semanitc segmentation network called deeplabv3 that had been trained on the cityscapes dataset. We used this trained network to get an image mask of just the sidewalk where we computed the centroid of the sidewalk and ran a gradient descent algorithm to follow the sidewalk.

Number 7

This is "lucky" number 7. This robot was built as the final project for the advanced robotics class at CU Boulder. The robot is a 1/10th scale remote control car with an odroid and lidar sensors. The goal was to make the robot race around a track created from a section of hallway in the engineering center. We had to contend with corners and large shipments of toiletries left in a section of the race track. This was my first real robotics project and it really was trial by fire. Thankfully I had an amazing team and I learned a lot from my team mates.

Co-Lab

Co-lab was a project for a software development class. The idea of the website was to combine trello with twitter or slack. Sharing information about projects and communicating effectively with remote members of work and tracking work projects. This project was more about learning how to use golang at the request of Rackspace before I started my internship there. Fun side note, I never wrote any golang during my summer at Rackspace, but I wrote a load of python widgets for dealing with databases.

MARBLE Gui

This is "The Gui", a human operator interface for DARPA SubT team MARBLE. The interface puts the operator in contact with the robots for sending and checking transforms, receiving and submitting artifacts found in the field, and controlling the robots when necessary. The bigger components I have added are around visialization of artifacts in rviz and being able to manipulate through rviz. I have also been working on cleaning up and fixing issues with the project as a whole. I have been removing unneeded dependencies, building and testing better setup scripts, organizing code, and updating the UI to better fit the operator's needs. As part of this project I left school for two weeks to drive the team truck full of robots out to Elma Washington where the DARPA SubT urban competition was held. It was two weeks of working around the clock and driving a large penske truck was tiring but it was one of the best experiences of my life.

Smartbases

The vehicle above is an electric golf cart with lidar, radar, gps, and cameras. This is the Smartbases project, a collaborative effort between CU Boulder, ERDC, FirstTransit, and us Ignite to research the deploment of autonomous vehicles for bussing around select Army bases. My parts of this project include working with the radar, communicating to computers on the base to gather traffic information, and interviewing people as part of a human factors study surrounding how people interact with the vehicle. My goal is to get radar SLAM working, to then be combined with lidar. I am aiming to use the radar data to track other agents in the world for the purpose of collision prediction. This project is particularly exciting for me as the saftey information we gain from this project will go on to save lives and improve autonomous vehicles for everyone.

Contact

Nikolaas.m.c.Bender@gmail.com

Click Here for mailto

Other ways to get in touch