library(MASS)
# rlm function does M-estimation
# by default it uses the Huber loss function
# you can specify others with the psi argument i.e. psi = psi.bisquare
# otherwise it works a lot like lm

x <- sort(4 * rnorm(20))
y <- 3 * x + c(rnorm(18), 32 * rnorm(2))
df <- data.frame(x = x, y = y)
plot(x, y)

fit.huber <- rlm(y ~ x, data = df)
coef(fit.huber)
abline(coef(fit.huber)[1], coef(fit.huber)[2])

fit.lm <- lm(y ~ x, data = df)
abline(coef(fit.lm)[1], coef(fit.lm)[2], lty = "dashed")

legend(-4, 80, legend = c("Least squares","Huber"), 
    lty = c("dashed", "solid"))
    

# Standardized residuals (just since you've seen them in class)
rstandard(fit.lm)

# what the book calls deletion residuals
rstudent(fit.lm)

# and some other things you may have seen
hatvalues(fit.lm)
cooks.distance(fit.lm)