Within Asimov's novels, there are two answers.
One is in the story Escape, which was included in I, Robot. I suspect this is the story that you half recall.
US Robotics feed the information on hyperspatial travel to the 'Brain', their positronic supercomputer. It is able to solve the problem and build a hyperdrive ship. The nature of the hyperspace jump is that the travellers arrive fine - but are dead during the jump. The way the problem is presented allows The Brain to solve the problem as they are ok in the end, but does cause something of a psychotic episode that Susan Calvin is confident they can treat.
'This so-called "death", in other words, was a strictly temporary
phenomenon. [...] Even with death temporary and its importance
depressed, it was enough to unbalance him very gently.'
Some robots, including Demerzel (Daneel), do become able to tolerate death or killing of humans, when they become advanced enough to appreciate the Zeroth law - that a robot may not harm humanity, or through inaction allow humanity to be harmed. This was first appreciated by Giskard in Robots and Empire, and is the scenario when robots are truly able to harm humans.