A carregar...
Publicação
RobotFix : Detecting Bugs On Variables In Robot Programs
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 1.08 MB | Adobe PDF |
Autores
Resumo(s)
Robotic systems are everywhere (e.g., factories or medical surgeries) and are widely
used with many real-world applications. However, programming robot systems is challenging, requiring domain-specific expertise.
The Robot Operating System (ROS) is a flexible framework for writing software for
large-scale robots, targeting a wide range of developers, from students to industry professionals. With ROS, developers can program robots to follow complex behaviors. However, due to its specificities, it is easy to introduce bugs in ROS projects through a typo
or making wrong assignments to variables regarding their unit or objective. Developers
need to detect these and then fix them, which takes time and reduces their productivity
and motivation. Furthermore, there is a lack of tools that help in detecting and fixing bugs,
as well as verifying the absence of domain-specific errors. Additionally, existing tools do
not provide the desired assurance on unit values and units. This work aims to improve the
robot programming process and help users make fewer mistakes while programming.
This work created RobotFix tool to help programmers in finding errors related to
variables. With RobotFix, developers can annotate the variable unit and intervals that
variable value should respect. With our approach, developers will know that they are
using the wrong unit (e.g., adding km/h and m/s) or a value that does not respect certain
conditions (e.g., a programmer introducing negative seconds into a variable).
This approach has the advantage of providing feedback about the correctness of the
code before executing, saving time and cost in testing the robot both in simulation or in
the field. Testing in robotics has its disadvantages, like the shortcomings of simulators and
the infeasible cost of testing all combinations of inputs. The approach was implemented
into a prototype called RobotFix. Then, I evaluated the approach on programs related to
20 bugs made by ROS programmers. RobotFix succeeded in finding the bugs and showed
that it could be a helpful tool for robotic programmers in the future.
Descrição
Tese de mestrado, Engenharia Informática, 2022, Universidade de Lisboa, Faculdade de Ciências
Palavras-chave
Robot Operating System Sistemas Robóticos Domain Specific Language Verificação estática Sistemas de tipos Teses de mestrado - 2023