Systematic software reuse in general is not an easy task and the robotic field is no exception. Yet some of the RSS projects are over seven years old, long enough to draw some conclusions. There has been undeniable progress in both software availability and in understanding in the community needs. But the result, from the point of view of day to day experience working with the robots have been underwhelming. The following problems laundry list is the result of subjective evaluation.
1. Slow growth; the main measure success of a project dedicated to software reuse in the reusable amount software it contains. In this respect, the projects of RSS show steady but growing slowly.
2. Fragmentation and duplication of efforts; very few developers contribute to more than a project. Basic robotic functionality implementation is replicated in each RSS project and is not reused across projects.
3. Poor software quality; hardware drivers are often of lack functionality and substandard reliability. Algorithms are aged and are not optimized for performance.
4. Software lock-in; software written to the API (application programming interfaces) which are unique to a particular RSS project is lock in. an organization or an individual who has contributed to the one project can not easily move their investment to another one. Ironically, this drawback is associated typically with proprietary rather than open source software.
5. Difficulty in making comparisons; numerous report have tried to compare different ways of building the systems of robotic, which in practice reduces to a comparison between RSS projects. Given the complexity of the task and fundamental differences in solution approaches, this inevitably leads to a comparison between “oranges and apples”.
Nothing indicates the above problems mentioned will not continue into the future. It mentions two more negative trends which are relatively new: danger from commercial alternatives and danger of irrelevance to the state of the robotic art.
No comments:
Post a Comment