Is there a way to launch the job again using Orchestrator, when an error occurs?
By current design, it is not possible to re-trigger the job when a fatal error is encountered using "Orchestrator Triggers". Refer Triggers to know more.
Handle errors within the process using some logic. For example, put a try catch block in the code and whenever a fatal error or any other exception occurs, trigger the same job again by using "Start Job activity". Refer Start Job .
Alternately, make use of "Queue triggers" providing a retry in the catch block. For instance, whenever an exception occurs, add an item in the queue and further the Queue Trigger triggers the same job again.
Refer About Triggers - Queue Triggers .