[ros-users] Looking for a ROS instructor/consultant to help in kicking off a new project

[Dwight Meglan]

We are starting an open source project for a surgical simulation system sponsored by the US Army.

Our company is composed of veterans of the surgsim world and has been developing a common surgical simulator core for a number of years now which has already resulted in several unique simulators to date, both with grant support as well as for private medical device companies. We've built up a nice core of real-time computational physics, haptics, display pipeline, metrics extraction, and so on. At the moment, we use both open source and closed source code.

The US Army is funding us to convert the core over to a fully open source system. After looking at various options, we plan to build off of ROS. The resulting system will have a BSD or Apache 2.0 license.

We've been using Constellation, a CASE environment and execution infrastructure developed by Real Time Innovations (rti.com) which spun out of the Stanford aero robotics lab some years ago. (They are known now for their distributed networking code, NDDS). It was originally designed for robotic systems and in going through the ROS tutorials, I can see conceptual similarities (and differences). Because of that approach, many parts of our simulator setup is already conceptually aligned to the ROS view of the world. But we need to do more redesign before we start adapting the code.

I've been through pretty much everything I could find on ros.org on learning ROS but there is a distinct gap for me still in moving from the component parts of ROS (which are clear and pretty familiar given our Constellation history) to the overall design and implementation at the system level.

So we'd like to find someone with considerable ROS system level experience to come work with us for a short period of time (2-3 days) to work us through the overall design tradeoffs in how to best go about splitting up and reconfiguring our simulator code. I looked about and did not find much in the way of prior postings by people having looked for similar expertise.

Note that I am aware of Gazebo and have looked at it to some degree. Our intent is to develop something not so monolithic and it will be specific to surgsim. We have built and use a number of custom haptic interface devices as well as camera and other tracking technologies so we feel ROS makes sense for us. But we have already striven to use data streams to make our rendering approach (using OpenSceneGraph) independent from the physics (our own though we have used many of the available physics codes before concluding we had to do our own) independent from the I/O devices, and so on. We are data stream oriented already and we'd like propagate this focus as deep into the system as makes sense. From what I have seen Gazebo mostly is linked to ROS at the I/O level. We hope to use ROS' infrastructure internal to the simulator itself.

I am glad to discuss design level issues, especially if my view of Gazebo is incorrect and/or if there are other sims out there that I should study.

As for instructor/designer opportunity, we are largely Boston area based and would bring you here for those days. We'd like to do this sooner than later. Early to mid September would be ideal.

thanks
--dwight

Dwight Meglan, PhD
Chief Technical Officer
SimQuest
dmeglan at simquest.com
www.simquest.com