ABSTRACT
This chapter deals only with errors in the environment. Recovery from environmental errors is far more complicated than recovery from simple controller errors, even expected errors can manifest themselves in unexpected ways. The interest in off-line programming makes automatic error recovery even more essential. The need for automatic and general-purpose error recovery will grow with the complexity of the applications and the software developed. Operating automatic error recovery techniques would avoid the need of identifying all the relevant cases, thus leading to a simplification of the programming techniques. Many ideas used to improve the reliability of robots have been taken from the fields of software reliability and software safety. A common approach to failure analysis and diagnosis is to apply techniques based on fault trees, event trees, or cause-consequent diagrams. Planning is a fundamental requirement of an error recovery system. Once an error has been detected and interpreted, the system must plan how the robot is to recover from the error.
