The implicitly and explicitly stated goal of several past and present standardization initiatives is to increase reuse software by enabling inter-operation between software from different RSS projects.
One possible solution is to achieve complete inter-operation by standardizing on a single way of building systems. While this approach may be appropriate for the military, mandates like this are not as effective in academics environments, especially when none of the existing solution is superior clearly. The multiple operating system existence, programming language and communication middleware packages also suggest that a one size fits all solution to building robotic systems may be undesirable and unachievable.
Another option is to enable communication between software from different RSS projects without modifying the individual projects. This can be done with translators custom designed for each application or with a more general software bridge. However custom solutions of this type tend to be too specific to be reusable and a generic bridge between middleware instances is a technical challenge. Beside, this approach requires extra effort during system operation as the user is asked to execute several systems and different types of infrastructure simultaneously.
It tries to show that diversity or lack of standards in approaches to building robotic system is not a problem itself. In fact it can be advantageous as long as it does not stand in the way of reuse software. We will argue that the bulk of the robotic software can be written in a way which makes it independent from the specific particular RSS and therefore reusable across multiple projects.
It expresses the approach in long term objectives for the open source robotic community:
1. Achieve a dramatic increase in the quantity and quality of available software which can be readily recombined for a particular application.
2. Maintain flexibility in how this recombination can be achieved.
No comments:
Post a Comment